Spectrum of the element being determinate; the acceptable spectral

dwidth varies between 0.1 and 1.0 nm. An unnecessarily large value will not
increase the amount of atomic absorption but may well result in increased
unwanted molecular absorption.
Filters provide adequate isolation of the required line if the element has a
very simple line spectrum. Thus filter photometers are frequently used for the
determination of sodium, potassium, calcium, and magnesium in clinical
samples.iron and nickel have more complex spectra and for these elements a
monochromator with a bandpass of about 0.2 nm is required to isolate individual
lines.
A photomultiplier is used as the detector, and for some elements the
photomultiplier needs to have a sensitive response in the UV region. The usefull
wavelength range is from 194 nm for arsenic to 852 nm for cesium. The overall
gain of the system is controlled by varying the voltage applied to the the
multiplier dynodes.
FLAME SPECTRA AND INTERFERENCES
Atomic Line overlap. Overlap of emission line may occur because of the
finite bandwidth of the wavelength selection system. Filter photometers are
particulary prone to this of interference, because of the relatively wide band of
wavelengths that they pass. The spectral bandwidth for a glass filter is typically
about 80 nm, and for an interference filter 15 nm. However, the effect also occurs
with monochromators. For example the magnesium line at 285.21 nm cannot be
isolated from the sodium line at 285.28 nm.
Until recently it was believed that this form of interference did not occur in
atomic absorption spectroscopy because a specific radiation source is used an the
absorption line are very narrow. Several examples have now been reported but
they are relatively rare in comparison with spectral interference effects in flame
emission spectroscopy.
Molecular emission and absorption. Flame emission spectra are
superimposed on a continous background radiation which arises from the flame
itself (see figure 13-2) or from molecular spesies present in the flame. Allowance
must be made for this in emission spectroscopy when the emission intensity from
the element of interest is relatively low. For example, the presene of large amount
of calcium interferes with the determination of small amounts of sodium and
barium because of the emission bands of calcium hydroxide in the vicinity of the
589- nm sodium and 554-nm barium lines.
The effect of flame emission pn absorption measurements is virtually
eliminated by modulating the radiation source. However, molecular absorption by
species such as strontium oxide and calcium hydroxide may interfere in atomic

absorption spectroscopy. The effect is generally more pronounced at low

wavelengths and high concentrations, and is reduced by the use of higher flame
temperatures. With a hollow-cathode line source, the wavelengths in the vicinity
of the resonance line cannot be scanned to obtain a background correction, but
correction can generally be made by measuring the effect of the moleculer
absorption on an otherwise nonabsorbing line close to the resonance line. An
example 231.7-nm nickel nonabsorbing line which can be used to correct for
nonatomic absorption at the 232.0-nm resonance line(see figure 13-11). A more
convenient method in the 190 to 320 nmrange is to use a hydrogen or deuterium
continuum lamp, in which case the line absorption and molecular absorption can
both be measured at the same nominal wavelength. This is possible because the
spectral slit width is much wider than the absorption line and the line absorption
due to the element of interest will have a negligible effect on the hydrogen or
deuterium continuum.
Physical effect and solution properties. Changes in the physical properties
of the aspirated solution change the absorption or emission intensity. The
viscosity of the solution afets the rate at which it is aspirated into the flame and
the surface tension, density, and viscosity of the solution, together with the gas
velocity, determine the droplet size. It is therefore important that the physical
properties of the sample and standard solutions should be matched as closely as
possible. The use of an organic solvent frequently increases the sensitivity by a
factor of 3 to 5. The amount of sample carried into the flame is increased because
of the lower viscosity compared with water. A reduction in droplet size also
occurs because of the lower surface tension and this increase the volatilization rate
as does a reduction in the solvent boiling point. Furthermore,with aqueous
solutions the large excess of water tends to interfere by forming nonemitting
radicals oe even oxides with some metals, and its molecular emission spectrum
may interfere with some emission measurements. Organic solvents reduce these
effects, although some water is still formed in the flame. They also increase the
reducing condutions in the flame.
Self-absorption. Part of the energy emitted from the interior of the flame
may be absorbed by atoms in the ground state in the outer regions of the flame. At
low sample concentrations this effect is negligible,but at high concentrations it
can be so pronounced that the emission intensity becomes nearly proportional to
the square root of the sample concentration. The effet is greatest when a long path
burner is used, as shown in figure 13-12. Self absorption may also occur with
hollow-chatode lamps if they are operated at too high a current. Unexcited metal
atoms in front of the chatodeabsorb radiationfrom the center of the resonance line,
and this distors the line profile. The distortion leads to reduced sensitivity and
more pronounced curvature of the calibration plots, as shown in figure 13-13.
However, the increased lamp emission may provide a better signal-to-noise ratio

with the acetylene-nitrous oxide flame in cases where a high flame background
emission produces an appreciable noise level in the photomultiplier.
Chemical effects and condensed phase interference. Chemical
interferenceoccurs as the result of formation in the flame of compounds that are
incompletely volatilized or dissociated. It is believed that most chemical
interference effects take place during the formation of solid particles in theflame
priorto their evaporation, and this is therefore often referred to as condensed phase
interference. An example is phosphorus interference in the determination of
calcium, which is due to the formation of calcium phosphate.
The effect of interferents can be established by measuring the emission or
absorption of a series of samples containing different concentrations of
interferences. In some cases the effect can be eliminated by changing the flame
conditions; for example, increasing the fuel flow rate to provide more reducing
conditions minimisize the formation of stable oxides. In others, a higher flame
temperature may be required and it is for this reason that the acetylene-nigrous
oxide flame is often preferred to the acetylene-air flame. However, it appears that
the flame chemistry is more important than flame temperature where the
formation of refractory oxides is concerned. Thus is has been found that refractory
compounds are poorly atomized in the hydrogen-nitrous oxide flame,although its
temperature is similar to that of the acetylene-nitrous oxide flame.this indicates
the need for carbon or carbon-containing radicals in the flame to minimize the
formation of metal oxides.
(Nexcess of a releasing agent can be used that will combine prefentially with the
interferent. An example of the latter is the addition of lanthanum or strontium
nitrate to calcium solutions to prevent the formation of calcium phosphate in the
air acetylene flame. A similar result can be achieved by protecting an element by
chelating it with EDTA. This forms a complex which protects the element from
unwanted reactions but is readily decomposed in the flame.
Flame temperature. The presence of other chemical compounds in the
flame affects both emission and absorption by changing the flame temperature. It
is obvious that any temperature change will affect the emission intensity, because
of the change in population of the upper energy state accordance with equation
13-1. The absorption will also change becausemost compounds are incompletely
dissociated, even in the acetylene-nitrous oxide flame, and the degree of
dissociation is therefore temperature dependent. Differentiating equation 13-1
with respect to temperature, and neglecting dissociation for the moment, we have
=(

exp (-E1/kT) = (N1/N0)E1/kT2

Since N0 is virtually independent of temperature (see table 13-2) we can rewrite

this as

dN1/dT =N1E1/kT2

(13-4)

consequently, a 1% change in the of atoms in the upper state (dN1/N1) is

equivalent to a temperature change T given by:
T = 0.01 kT2/E1

(13-5)

T = 7.0 x 10-10 (nm) T2

( 13-6)

For a flame temperature of 3000 K, and transition which produces a line at 300
nm (equivalent to 4 eV), this coreesponds to temperature change of about 2 K.if
the dissociation energy is also of the order of 4 eV, we can expect a temperature
change of 2 K to produce a change in absorption of 1% in emission intensity 2%.
Ionization. As hs al ready been discussed, pronounced ionization of
elements with low ionizationpotentials occurs in high-temperature flames.
Because the extent of ionization decreases with increased metal concentration in
the flame, an upward curvature of the calibration plot results. As shown in figure
13-14, the effect can be overcome by adding an excess of a more readily ionized
element. This creates high electron concentration and suppresses the ionization of
the less easily ionized element. A relatively high concentration of potassium, in
the order of 2000 g/ml, is generally used for this purpose.

QUANTITATIVE ANALYSIS
Sample Preparation. The sample pretreatment required depends on the
element to be determined. The substrate material, and the method of stomization.
With the graphite furnance most samples require no chemical pretreatment
because the chemical matrix is removed by ashing to atomization. For flame
atomization most liquid samples can be sprayed directly into the flame after
dilution with a suitable solvent. Solid samples sre generally dissolved in acid but
some may require alkali fusion before acid treatment.hydrocloric, nitric, an
sulfuric acids are commonly used for the dissolution of metals and alloy. Nitric
acid gives compounds that are relatively easily decomposed but not higly volatile,
and is therefore used fore ashing in preference to hydrochloric acid. A 3:1:1
mixture of nitric, sulfuric, and perchloric acids is useful for the wet oxidation of
organic matter. ( caution : percholic acid and perchlorates may be unstable, and
when mixed with carbonaceous materials, form explosive mixtures. See Christian
and Feldman, p. 193, concerning the use of this oxidizing mixture). Note that
reagent grade acids can contain significant amounts of trace metal impuritieintos;
chromium in nitric acid, lead in hydrochloric acid and cadmium in sulfuric acid
are particularly common.
An organic solvent may be used to selectively extract metals after
complexing from an aqueous solution, and the solvent can then be aspirated

directly into the flame. Methyl isobutyl ketone an ethyl acetate are preferred for
this purpose because of their suitability as extractive agents and for use in premix
burner systems. Detailed method sheets for the analysis of a wide variety of
samples by atomic absorption spectroscopy are aviable from a number of
instrument manufacturers.
Standards. Standards and sample solutions should be as similar as
possible, both in chemical and physical properties. Reagents need to be checked to
unsure that they do not introduce significant amounts of the element tobe
determined. Standards and samples should be stored only in polyethylene bottles
because some metals adsorb onto glass surfaces. Standards containing low
concentrations to avoid adsorption problems.
Calibration techniques. The calibaration technique which is most
appropriate for a particular analysis depends on the number of samples to be
analyzed, the linearity of the calibration graph, and the extent to which other
components in the sample interfere with the analysis.
Working curves. if a number of samples are tobe analyzed, the simplest
procedure is often to prepare a series of standards covering the concentration
range of interest and from these obtain a working curve as shown in figure 1315. This tends to be nonlinear, but devices are commercially available that will
electronically compensate for the curvature and thereby enable a direct readout
linear in concentration to be obtained sometimes an alternative wavelength can be
selected that yields a more linear calibration curve. For example, the 235,4 or
286.3-nm absorption line for tin provides a more linear response that the 224.6nm line which is very close to an interfering line.
Neglecting interference effects, flame emission provides a signal which
is directly proportional to concentration. Atomic absorption, however, follows
beers law because an absorption process in involved. In the latter case the
amplifier gain is set to provide a signal corresponding to 100% transmission when
a blank solution is aspirated into the flame. The corresponding absorbance for
each standard is then given by
A= - log T =2.000 (% T)

(13-7)

Which follows directly from the use of beers law. Alternatively, the % T
calibration data can be plotted directly on semi-logarithmic paper. With most
modern instruments the absorbance is determined electronically by means of a
logarithmic function generator, and can be read out directly.
The working curve must be checked from time to time, since the slope of
the curve may change due to minor fluctuations in the fuel ND OXIDANT
PRESSURES or in the sample floe rate. Appropriate adjustments can then be
made to comply with the previously determined calibration data. The

measurements will almost always be made at the wavelength corresponding to the

emission or bsorption maximum, because this is least aaffected by a small shift in
the wavelength setting such as may result from a change in tempearature of the
monochromator. It should also be remembered that emission and absorption
varies eith the position of the burner, and for any on element the emission and
absorption maxima(see figure 13-6) rarely occur at the same height in the flame.
In emission spectroscopy there is generally a significant contribution to
the signal due to the flame background and this may need to be subtracted from
the measured signal intensity. Thi. If s can be done by measuring the background
signal close to the emission line used for determination of the test element. If the
background varies with wavelength, an average value for the background at the
line center must be obtanined from measurements on ethier side of the line. Any
variation will be obvious with a recording instrument which scans across the line
but with a nonrecording instrument it may be necessary to take three separate
measurements, one at the center of the line and one on ethier sid of it. It is
obviously necessary to unsure that the same three wavelengths are used for eah
separate determination. In atomic absorption, particularly at wavelengths below
220 nm. In that case th background absorption is determined at the line center by
using a hydrogen or deuterium continuum lamp.
Internal standardization. Minor variations in the fuel supply or in the
physical properties of the sample may be compensated for by adding a fixed
amount of an internal standart to each sample. The calibration graph is then a plot
of the ratio of the emission intensity, or the absorance, of the element being
determined to.