Optical Atomic Emission Spectrometry/Flame Photometry☆

María Luisa Fernández-Sánchez, María Teresa Fernández-Arguelles, and José Manuel Costa-Fernández, University of Oviedo,
Oviedo, Spain
© 2018 Elsevier Inc. All rights reserved.

Introduction 1
Principles of Flame Optical Emission Spectrometry 1
Flame Characteristics 3
Instrumentation 4
The Burner 5
Wavelength Selection and Detection Systems 5
Performance Characteristic 5
Linearity of Response 5
Detection Limit 5
Precision 5
Interference Effects 6
Physical effects 6
Chemical effects 6
Ionization interferences 7
Spectral interferences 7
Practical Applications of Flame Photometry 7
Further Reading 9

Introduction

The history of atomic spectroscopy is closely associated with the use of flames as light sources since more than 200 years ago. The
systematic attribution of spectra to chemical elements began in the 1860s when the scientists Bunsen and Kirchhoff showed that the
radiation emitted from the flames depends on the characteristic element present in the flame. Qualitative analyses for several
elements were being carried out with this tool over the last 100 and 50 years. Quantitative analyses evolved since 1930s, as
spectrometers and light-measuring devices were improved. The peak of interest and use of flame atomic emission was reached at
about 1960. Thereafter atomic absorption took over much of the field (absorption measurements provide to be more reliable and
the range of elements can be determined is broader).
Graphite furnace (GF)-based methods come afterwards and extended then the range of usefulness of atomic absorption (much
better sensitivity than with flame) when the somewhat more complex needed instrumentation was quickly put in a more
reliable form.
More recently, high-energy plasmas with very sophisticated instrumentation have further replaced flames as spectrochemical
sources for optical atomic emission.
In this article the basic processes of flame emission will be described and its present usefulness re-examined. In spite of being the
first used, simpler source, modern FES developments have rendered flame photometry a useful analytical tool for elemental analysis
for some selected elements even today.

Principles of Flame Optical Emission Spectrometry

In flame optical emission spectrometry (FOES), the sample solution is converted into a fine aerosol via a nebulizer and introduced
into the flame. The heat from the flame evaporates the solvent and breaks chemical bonds to create free atoms. The thermal energy
also excites the created atoms up to excited electronic states that subsequently emit light when the excited electron returns to the
ground electronic state. Each element emits light at a characteristic wavelength, which is dispersed by a grating or prism, while a
detector, normally a photomultiplier, is used to convert the emission into a measurable electrical signal.
The distribution of atoms of an element among available states of energy (Ei) above the ground state (E0) is governed by their
energy difference, DEi, and the absolute temperature of the source according with the Boltzmann equation:

☆
Change History: October 2018. María L Fernández-Sánchez, María T Fernández-Arguelles, and José M Costa-Fernández, updated the text and Further Reading of
this entire article, and they also added the Abstract, Keywords, and some of the sections, Figure 1 and Figure 3. Renumbered Figure 1, Figure 2, Figure 3, Table 1
and Table 2.

Encyclopedia of Analytical Science, 3rd Edition https://doi.org/10.1016/B978-0-12-409547-2.14533-0 1

2 Optical Atomic Emission Spectrometry/Flame Photometry

Ni =N0 ¼ ðgi =g0 Þ eDEi =kT (1)

where Ni is the number of atoms with Ei energy, N0 the number of atoms in the ground state, gi and g0 describing the degeneracy of
each energy level, D Ei is the energy difference between of the excited state (Ei) and the ground state (E0 ¼ 0), k is the Boltzmann’s
constant (1.3807  10–23 J K1), and T is the absolute temperature (Kelvin degrees).
Compared with other common excitation sources- such as the arc, sparks and plasmas- flames have rather low temperature, up
to 3000 K. Most atomic transitions occur then between the ground state and a few low-lying excited states, even with elements
having great numbers of possible energy states. Table 1 gives the values of the Boltzmann factor (eDEi/kT), at known intervals
throughout the spectrum, for 2000 K and 3000 K, the useful flame temperature range. When corrected for their multiplicity (small
whole number factors, gi/g0), the values give the fractions in the excited states directly (see Eq. 1) for a given T and a given atomic
line (DEi.). Due to the relatively low temperatures in flames, it is apparent that the vast majority of atoms in flames are in their
ground states.
The intensity of an atomic emission line, Ie, is proportional to the number of atoms, populating the excited state, Ni, in a given
volume of the flame observed

Ie ¼ kNi (2)

where k is a constant accounting for the efficiency of the transition.

By substituting the value of Ni, on the intensity equation:

Ie ¼ kN0 ðgi =g0 Þ eDEi =kT (3)

From Eq. (3), we expect that excited states with lower energies (smaller DEi) will have larger atom populations. The most intense
emission lines, often called “resonance lines,” are those arising from permitted transitions between the ground state and certain low-
energy excited states of course, we also expect emission intensity to increase with the temperature of the source.
Under flame controlled conditions then, the number of atoms per unit volume in the ground state (N0) will be approximately
equal to the total number of atoms per unit volume of the analyte atoms in the sample which is actually introduced into the flame.
Therefore, if the concentration of the element to analyze in the solution is C (mol L1), the intensity of the emission constant
conditions will be given by

Ie ¼ KC (4)
thus, the intensity of the emitted light is related with the element concentration dissolved sample introduced into the flame (C) (K is
the constant of proportionality).
Those basic considerations have some practical conclusions:

1. In hotter flames, elements having resonance lines at longer wavelengths may show greater sensitivity (lower detection limits) by
emission than by atomic absorption. In an absorption measurement, we must always measure the ratios between the incident
and the transmitted light intensities (the detection limit is reached when the difference between both values approaches the
combined possible measurement variations in these two intensities). Flame emission relies on a single measurement or the
difference between sample and blanks signals has no such restraint.
2. Flame emission spectra are relatively simple. Thus, simple and small spectrometers are adequate. Of course, more versatile and
larger modern instruments have greater resolving power so that more spectral lines can be resolved and the background
continuum can be made weaker (improved analytical performance).
3. Spectral interferences can occur and excited species may be produced by mechanisms other than thermal excitation
(e.g., chemiluminescence). Thus, the measured intensity is greatly increased and possible errors should be taken into account.

Table 1 Boltzmann factors for the practical range of resonance line wavelengths
and flame temperatures (T )

Boltzmann factors, eDEi/kT

Wavelength of resonance line (nm) 2000 K 3000 K

200 2.44  1016 3.89  1011

300 3.89  1011 1.15  107
400 1.55  108 6.27  106
500 5.69  107 6.84  105
600 6.27  106 3.39  104
700 3.46  105 1.06  103
800 1.25  104 2.50  103
900 3.39  104 4.8  103
 Optical Atomic Emission Spectrometry/Flame Photometry 3

Flame Characteristics

The most widely used flame in spectrochemical analysis is the laminar premixed flame of a gaseous fuel (e.g., C2H2) and the
oxidant, usually air. It is the simplest flame and exhibits characteristics common to many other systems. Fig. 1 delineates the main
regions of the flame:

 The primary reaction zone is the ignition zone, in which a partial combustion takes place. The reactions occurred in this zone are
not in thermal equilibrium and the concentration of free radicals is very high. Since a great emission of light is observed, this zone
is not used in flame spectrometry (too high backgrounds).
 The interconal zone or interzonal zone, situated immediately above the primary combustion zone. The flame gases achieve the
maximum temperature there as a complete combustion takes place and almost a thermodynamic equilibrium is reached. This
area or zone of the flame is rich in free atoms and it is the most widely used part of the flame for analytical spectroscopy.
 The secondary reaction zone (see Fig. 1), where the products of the combustion processes, in contact with surrounding air, are
converted into stable molecular species.

The temperature and the maximum burning velocity of several common analytically useful flames are given in Table 2. These are
so-called theoretical temperatures, calculated for stoichiometric mixtures of fuel-oxidant gas by Snelleman. They are roughly one
hundred degrees higher than most measured temperatures. Moreover, the stoichiometric mixtures do not give the highest attainable
temperatures: these are reached at somewhat higher fuel-to-oxidant ratios, especially for the air-acetylene flame (due at least in part
to air entrainment). Fuel richness also alters reaction rates and extents of chemical reactions in flames. In any case, the tabulated
values in the Table show the relative temperatures of useful flames. The flame burning velocities determine the maximum gas flow
rates needed to keep the flame burning smoothly above the burning slot (Table 2). The residence time, during which a given atom
passes through the optically observed flame region, is thereby established— 1 ms in common flames. The slower burning
nitrous oxide-acetylene flame is thus preferred to the hotter but much faster-burning oxy-acetylene (O2–C2H2) flame, which is
also inclined to explode.
Due to the higher T, the hotter nitrous acetylene flame has a more complex and intense spectrum than the air acetylene flame.
The spectrum of nitrous oxide-acetylene flame when water is nebulized showed several species that emit radiation, such as C2, CH,
OH, and CN (Fig. 2). The background emission observed is from both the secondary combustion zone and the interzonal region.
This phenomenon (background emission) does not represent a major drawback in absorption measurements but in emission
measurements care must be taken to avoid overlaps with the emission lines of the analytes. It should be stressed how the flame
background emission can be limiting factor in determining the precision and detection limit of a given element determination.

Secondary
reaction zone

Interconal zone

Primary
reaction zone

Premixed
gases and
fine droplets
of solution
Fig. 1 Zones of the flame.

Table 2 Theoretical flame temperatures and maximum burning and velocities of some useful flame

Flame Volume ratio Temperature (K) Maximum burning velocity (cm s1)

Air–C2H2 12.5 2537 158–266

Air–C3H8 25 2267 39–43
O2–C2H2 2.5 3343 1100–2480
O2–C3H8 5 3094 370–390
N2O–C2H2 5 3148 285

Reprinted from Alkemade, C.Th.J.; Herrmann, R. FundaMentals of Analytical Flame Spectroscopy (trans. Auerback, R. and Gilbert, P.T. Jr.). 1979; New York: Halsted Press.
4 Optical Atomic Emission Spectrometry/Flame Photometry

Fig. 2 Emission spectrum of a nitrous oxide-acetylene flame, somewhat rich in C2H2 (red cone 3 mm high above the burner). Scan rate 12.5 nm/min, 25 mm slits.
The principal molecular emission bands are labeled.

Instrumentation

The basic components of a flame emission spectrometric set-up provide the means to achieve the following functions: deliver the
sample to the flame, induce thermally the spectral electronic transitions necessary for the final determination of the analyte, isolate
the spectral lines required for the analysis, detect the increase or decrease in intensity of radiation at the isolated lines and record the
intensity data observed.
The general layout of components for a flame emission spectrophotometer is shown in Fig. 3.

Fig. 3 Schematic diagram of a flame spectrophotometer. Adapted from File:Flame emission schematic.png. Wikimedia Commons https://commons.wikimedia.
org/wiki/File:Flame_emission_schematic.png
 Optical Atomic Emission Spectrometry/Flame Photometry 5

The Burner
The burners used as excitation sources in analytical flame spectroscopy should produce a stable and reproducible flame supplied
with the analytical sample solution. There are two types of aspirator-burner used, total-consumption burner and premix chamber
burner.
In a total consumption burner, both gases (fuel and oxidant) and the sample, come together at the exit of the burner at the base
of the flame. The sample is aspirated through a capillary by Venturi effect created by the flow of gases around the tip of the capillary.
These burners have the advantage that the entire sample is introduced into the flame.
In any case, all more modern commercial flame instruments use a premixed burner system based on three components (see
Fig. 3): (1) a nebulizer that breaks up the liquid into only small droplets; (2) a spray chamber, where the generated aerosol is mixed
with the fuel and flows through the “aerosol modifiers,” that drained large droplets from the stream, allowing only droplets (smaller
than a certain size to pass), and (3) the flame or atomizer that converts the analyte into free atoms and/or excited atoms via thermal
energy.
Two important features of this burner are that only a small amount (5%–10%) of the total liquid sample aspirated reaches the
flame in the form of small drops, and that a relatively low noise flame is achieved. Both features (together with its relatively long
optical path) increase the signal, the analytical sensitivity and the reproducibility of the measurement.

Wavelength Selection and Detection Systems

For routine flame emission determinations of alkali and alkali-earth elements, interference filter photometers are often used (i.e.,
having sufficient resolution) to separate the appropriate emission line selected, while a light transducer, sensitive to light, is used as
detector. Some flame photometers have been specially designed for the analysis of lithium, sodium and potassium in biological
samples. There, the radiation from the flame is split into three light beams, each of which passes through the corresponding
interference filter, then a photocell detects the emitted light in each channel and converts it into a voltage to be recorded and
measured. The three output signals can be measured independently, although, in general, lithium signal serves as the internal
standard.
Likewise, grating monochromators, controlled by a computer, are most often used for the rapid and sequential measurement of
the radiation emitted at different wavelengths (see Fig. 3), and high-sensitivity multichannel detectors, such as linear silicon
photodiode array (PDA), are mostly used as light detectors when sequential and simultaneous multielement detection are desired.

Performance Characteristic
Linearity of Response
The intensity of the flame emission observed varies linearly in orders of magnitude with the analyte (atom) concentration (see
Eq. 4), with the upper limit set by a curvature of the calibration curve due to self-absorption. With the 5 cm slot burner, linearity is
good below 1 mg L1 for most elements and nearly perfect below 0.1 mg L1. For potassium this means four orders of magnitude of
linearity and for lithium five orders. The upper limit may be modified by setting the slot burner cross-wise (shorter light path-
length), if needed.

Detection Limit
The detection limits reported for different element by flame emission and atomic absorption, with different type of flames, are given
in Table 3 for comparison. Flame richness (higher C2H2 content) greatly influences background intensity and, for many elements,
line intensity as well, generally with both increasing together but at different rates. The strong oxide forming elements, denoted by R
in Table 3, are more sensitive in rich red cone of nitrous oxide/acetylene flames (more reducing), by as much as 50-fold for certain
rare-earth elements. For these, the red cone (CN molecular emission) needs to be increased to at least 1cm flame height above the
burner. For nonoxide formers (L in the Table), maximum intensity is obtained with an oxidizing nitrous oxide-acetylene flame (red
cone 1 mm high above the burner). Other elements (I) are best determined with 3–5 mm high red cone flame. The required height
of the red cone is usually specified in the handbooks of standard conditions The comparison of both techniques shows that flame
photometry emission is more sensitive than flame AA when the excitation potential of analytical line is lower than 3.5 eV
(>350 nm) and providing better detection limits (e.g., for alkali-earth elements).

Precision
For samples and standards in which the analyte concentration exceeds the detection limit by at least a factor of 50, the relative
standard deviation of measurements in the flame is about 1%–5%. Perhaps the most important factor affecting precision is the
stability of the flame’s temperature. For example, in a 2500 K flame with a temperature fluctuation of 2.5 K gives a relative
standard deviation of 1% in emission intensity. Significant improvements in precision may be realized by using adequate internal
standards.
6 Optical Atomic Emission Spectrometry/Flame Photometry

Table 3 Detection limits by flame emission and atomic absorption spectrometry

Emission Absorption

Element Wavelength (nm) N2O–C2H2 Air–C2H2 N2O–C2H2 Air–C2H2

Al 396.1 0.02L
309.2 0.1
Ba 553.6 0.001L 0.05
Ca 422.7 0.001L 0.005 0.002
Cr 425.4 0.005I
357.9 0.005
Dy 404.6 0.07R
421.2 0.2
Er 400.8 0.04R 0.1
Eu 459.4 0.001R 0.08
Fe 372.0 0.05L
248.3 0.005
Ga 403.3 0.01L
287.4 0.07
Ho 405.4 0.02R
410.4 0.1
ln 451.1 0.002L
303.9 0.05
K 766.5 0.0001 0.005
Li 670.8 0.00001L 0.0001 0.005
Mg 285.2 0.005L 0.0003
Mn 403.1 0.005I 0.0003
279.5 0.002
Na 589.0 <0.0005 0.002
Rb 780.0 0.0002 0.005
Ru 372.8 0.02
349.9 0.3
Sc 402.0 0.03R
391.2 0.1
Tl 535.0 0.02L
276.8 0.2
Tm 371.8 0.02R
410.6 1
Yb 398.8 0.002R 0.04

R, strong oxide-forming elements; L, monoxide-forming elements; I, other elements.

Concentrations, in mg L1, giving signals twice the r.m.s. noise in the flame background.

Interference Effects
Both chemical and physical effects can cause line intensities for samples to differ for those for simple standard solutions at the same
analyte concentration. The interferences more common in flame spectroscopy can been classified, in terms of with their mechanism
of origin, as: physical, chemical, ionization and spectral interferences.

Physical effects
Several physical properties associated with the dissolved sample may need to be controlled including: solution temperature,
viscosity, surface tension and vapor pressure. They influence nebulization and transport efficiency of the sample to the flame.
Many of such effects are interdependent and not easily separated for individual effect For example, enhancement of signals by
addition of flammable organic solvents (i.e., ethanol) was common practice in the past. This caused changes in aerosol formation,
in transport, in droplet size distribution, evaporation rates, and flame temperature. Although the solvents decreased the flame’s
temperature by partially replacing higher energy acetylene fuel, other processes were accelerated. Stability generally worsened, and
signals were often slow to reach their final average values. Organic solvents today are most often used just incidentally (e.g., in the
solvent extraction of organic chelates of the analytes) to enrich and separate them from interferences.

Chemical effects
Chemical interferences arise whenever analyte combine with other sample constituents, at the temperature of the flame, to alter
their rates of vaporization compared with those of standards. The most common cause of this type of interference is the presence in
the sample of anions that form low volatile compounds with the analyte (reducing in this way the degree of atomization) A classic
 Optical Atomic Emission Spectrometry/Flame Photometry 7

example is the error caused by PO3

4 in the atomic emission of Ca, since it forms calcium phosphate, a refractory salt to dissociation
at the relatively low temperature of the flame.
Some useful methods of overcoming these and other major interference effects in FOES include:
1. The use of standard additions: each sample is used for preparing its own calibration graph, but this is very slow.
2. Add a “release agent,” a salt of another metal added in large amount to displace the analyte forming less refractory compounds
(lanthanum, at 100-fold or higher concentration than the analyte, is most often used).
3. Go to a hotter flame. The calcium-phosphate effect virtually disappears on a hotter nitrous oxide- acetylene flame.
4. Investigate flame richness. Effects for several elements are diminished in leaner flames. This cannot be used for analytes requiring
rich flames (e.g., rare-earth elements).
5. Separate analyte from matrix. This is most helpful when it is also used to concentrate or enrich the analyte, that is, to transfer it to
a smaller volume of solution (perhaps along with others analytes at the same time).
6. Add a bulkier, more volatile matrix. Certain high boiling point acids, and many organic compounds that decompose without
melting when heated, have been found to suppress interferences from several elements in flames even at concentrations as low as
a few milligrams per liter. Rare–earth elements show this effects most strongly, but several common transition elements also do
(e.g., cobalt and nickel).

Ionization interferences
Ionization interferences occur when thermal energy from the flame, is sufficient to ionize the neutral atoms of the analyte, M:

MðgÞ Ð Mþ ðgÞ þ e

Any ionization reduces the population of both the ground state and the excited state of neutral free atoms, thus lowering the
sensitivity of the determination. In hot flames, such interference is encountered with easily ionized elements, such as alkali metals
and alkaline earths. Manganese, with a fairly typical ionization potential of 7.4 eV, is 10% ionized in a nitrous oxide-acetylene
flame. Alkali metal salts are added at a few hundred times greater concentration than analytes to suppress ionization. The same
amounts are added to standards. The addition of suppressants is even more important in analyses that require a hotter acetylene/
nitrous oxide flame (errors due to ionization may easily be reduced to 1% or 2%).

Spectral interferences
Overlapping of the analytical wavelength selected with that of others emissions (e.g., of the sample matrix or other element) can
originate positive errors. Spectral interferences are observed when the continuous background radiation emitted by the flame occurs
at the same wavelength of the analyte. In addition, the sample matrix can also contribute to background. The molecular spectra of
metal oxides are essentially continuous, even using a high resolving power spectrometer. The resonance lines of barium at 553.6 nm
and sodium at 588.9 nm suffer from serious spectral interferences by the strong band emission of CaO that spreads over several
hundred nanometers.
This type of interference is important in any optical emission based technique.
With rather unsophisticated instruments, background can be corrected by scanning or successive online and offline measure-
ments. In general, scanning is needed when background is intense whereas the other simpler method is enough to correct a weak
background.

a. Scanning: The spectrum is scanned and recorded across the complete line profile at 0.2 nm per min for perhaps 0.4 nm overall
(Fig. 4). The height of the lithium peak above the small CaO fine structure peak on right is to be compared with that of the
calcium standard scan. The correction is accurate within 2%, even though the Ca:Li ratio is 1.6  106 in the samples.
b. Two-or-three wavelength method: Fig. 5 illustrates the procedures for the most general case, when the flame background and
sample background both change with wavelength close to the line. The corrected line intensity is obtained by subtracting the
average IB1 and IB2 from ILþB. If it is known previously that the background is flat, only one offline reading is needed. The
readings should be made at wavelengths removed from the line by perhaps five times the spectral bandpass, where the line
profile virtually disappears. The operator must always study the background problem whenever an unfamiliar element–sample
combination appears (the two–wavelength method is often required, but it is easily accomplished).

Practical Applications of Flame Photometry

Most applications of optical atomic spectroscopic techniques have been devoted to the determination of trace metals, especially in
liquid samples. Prior to the analysis, solid samples are generally solubilized by an appropriate digestion or leaching in oxidizing
acidic mixtures (aided by heat, ultrasound or microwave radiation). It is important to note that the resulting solution must be free of
particles that build up and block the nebulizer. Also, samples with a high saline content can cause interferences and deposits on the
burner and should be avoided.
At present flame emission still retains advantages in approximately the same area in which it began, that is the determination of
alkali metals. Flame emission is now used in clinical laboratories to determine the levels of sodium and potassium in blood,
8 Optical Atomic Emission Spectrometry/Flame Photometry

Fig. 4 Four 0.3 nm scans of the lithium 670.8 nm line; scan speed 0.2 nm min1. (A) nitrous oxide–acetylene flame only; (B) 2 mg L1 lithium standard solution;
(C) 3.2 g L1 calcium, as purified CaCO3 in 0.1 mol L1 HCl, with 2 mg L1 lithium added; (D) CaCO3 solution only, same concentration as in (C). All are actual
recordings, starting, and ending with zero light intensity (beam blocked), showing low electrical noise.

Fig. 5 Diagram illustrating the three-wavelength background correction method, when both flame and sample background contributions vary near the line. Online
reading in the center; readings offline by 0.2 nm, one on each side of line. Zero intensity is at the low end of the axis. Each trace stars and ends with flame
background intensity.
 Optical Atomic Emission Spectrometry/Flame Photometry 9

plasma, urine, and cerebrospinal fluid (CSF). Analyte concentrations and their inherent line intensities are such that samples must
be greatly diluted. Doing so various difficulties can be avoided. The use of lithium salts, added at a high concentration as internal
standard, helps to correct for some of the uncontrolled variables. The widely used clinical three channel filter photometer, designed
to determine Na, K, and Li simultaneously, has been described above in Instrumentation.
Flame emission has also shown its suitability to determine the levels of Na in extracellular fluids and K in intracellular fluid and
the amount of Li administered for therapeutic purposes (e.g., bipolar disorders) in biological fluids.
Apart from determinations alkali metals, Table 3 suggests some others applications, especially when analysts need to avoid the
high costs of modern plasma methods. With few exceptions, all elements listed there give at least three times lower detection limits
by flame atomic emission than by flame atomic absorption. This may be considered sensitive enough, with detection limits below
0.1 mg L1. The four exceptions are chromium, iron, magnesium, and manganese. The great sensitivities for aluminium, calcium,
strontium and barium by emission should be valuable in geological and agronomic work.
In brief, flame optical atomic emission offers a very simple, inexpensive, and sensitive method for detecting alkaline and alkaline
earths, as well as several transition metals including Fe, Mn, Cu, and Zn. The simplicity of this technique has found wide
applications in clinical, agricultural and environmental analysis, also for industrial analyses of ferrous metals and alloys, ceramic
materials, cements, foods, beverages, and petroleum products.

Further Reading

Cresser, M. Flame Spectrometry in Environmental Chemical Analysis: A Practical Guide. Spectroscopy Monograph Series; Royal Society of Chemistry: Cambridge, 1994.
Robinson, J. W.; Skelly Frame, E. M.; Frame, G. M., II Undergraduate Instrumental Analysis, 7th edn; CRC Press: New York, 2014.
Ingle, J. D., Jr.; Crouch, S. R. Spectrochemical Analysis; Prentice-Hall: Englewood Cliffs, NJ, 1988.
Jackson, K.; Wand Chen, G. Atomic-Absorption, Atomic Emission, and Flame Emission Spectrometry. Anal. Chem. 1996, 68 (12), 231R–256R.
Dean, J. A., Rains, T. C., Eds.; Flame Emission and Atomic Absorption Spectrometry. In Theory; Dekker: New York, 1969; vol. 1.
Dean, J. A., Rains, T. C., Eds.; Flame Emission and Atomic Absorption Spectrometry. In Instruments and Techniques; Dekker: New York, 1971; vol. 2.
Dean, J. A., Rains, T. C., Eds.; Flame Emission and Atomic Absorption Spectrometry. In Elements and Matrices; Dekker: New York, 1975; vol. 3.
Skoog, D. A.; Holler, F. J.; Crouch, S. R. Principles of Instrumental Analysis, 7th edn; Cengage Learning: Boston, 2017.
Pickett, E. E.; Hawkins, J. L. Determination of Lithium in Small Animal Tissues at Physiological Levels By Flame-Emission Photometry. Anal. Biochem. 1981, 112, 213–218.
Neel, B.; Crespo, G. A.; Didier Perret, D.; Cherubini, T.; Bakker, E. Camping Burner-Based Flame Emission Spectrometer for Classroom Demonstrations Bastien. J. Chem. Educ. 2014,
91, 1655–1660.