Designation: E 158 – 86 (Reapproved 1996)e1

Standard Practice for

Fundamental Calculations to Convert Intensities into
Concentrations in Optical Emission Spectrochemical
Analysis1
This standard is issued under the fixed designation E 158; the number immediately following the designation indicates the year of
original adoption or, in the case of revision, the year of last revision. A number in parentheses indicates the year of last reapproval. A
superscript epsilon (e) indicates an editorial change since the last revision or reapproval.

e1 NOTE—Keywords were added editorially in June 1996.

1. Scope Plotting Analytical Curves with a Variable Internal Stan- 7

dard
1.1 This practice covers fundamental forms of plots, used to Background Correction 8
convert measured intensities into concentrations in both pho- Concentration Calculations 9
tographic and direct reading analysis. It includes equations Corrections for Interferences from Other Elements 10
suitable for use with calculators or computers for such conver- 1.5 This standard does not purport to address all of the
sions, and methods for making background corrections and safety concerns, if any, associated with its use. It is the
corrections for interferences from other elements. responsibility of the user of this standard to establish appro-
1.2 A previous issue of this practice attempted to treat priate safety and health practices and determine the applica-
optical emission and X-ray fluorescence as one field, employ- bility of regulatory limitations prior to use.
ing similar calculations to convert measured radiation intensi-
ties into concentrations. In fact, quite different procedures are 2. Referenced Documents
used in the two fields. This practice confines itself to optical 2.1 ASTM Standards:
emission spectrochemical analysis. A separate practice will be E 116 Practice for Photographic Photometry in Spectro-
devoted to calculations in X-ray fluorescence spectroscopy. chemical Analysis2
1.3 This practice is related to but should be distinguished E 135 Terminology Relating to Analytical Chemistry for
from: Metals, Ores, and Related Materials2
1.3.1 Practice E 116, which describes procedures for con- E 305 Practice for Establishing and Controlling Spectro-
verting either transmittance or optical density of lines on a chemical Analytical Curves2
photographic emulsion into intensities or relative intensities. E 876 Practice for the Use of Statistics in the Evaluation of
This step is required in photographic analysis before the Spectrometric Data3
present practice can be applied.
1.3.2 Practice E 305, which describes the procedures for 3. Terminology
fitting a curve to a group of plotted points, and for standard- 3.1 For definitions of terms used in this practice, refer to
izing the equipment, in order to obtain optimum accuracy, Terminology E 135.
guided by sound statistical practice.
1.3.3 Practice E 876 discusses the statistical treatment com- 4. Significance and Use
monly applied to spectrochemical data, such as calculating the 4.1 Over the years, many workers have developed improved
standard deviation and the relative standard deviation. methods for plotting the measured intensities obtained from
1.4 This practice includes the following sections: chemically analyzed specimens (“standards”), and for calcu-
Section lating concentrations of unknowns from the “analytical” curves
Referenced Documents 2 obtained. This practice describes the methods of plotting and
Terminology 3
Significance and Use 4 the calculation techniques which have been found to produce
Fundamental Theory 5 superior results.
Plotting Analytical Curves with a Constant Internal Stan- 6
dard 5. Fundamental Theory
5.1 In optical emission spectrochemical analysis, there are
relatively few interferences from other elements, for materials
1
This practice is under the jurisdiction of ASTM Committee E-1 on Analytical
Chemistry for Metals, Ores and Related Materials and is the direct responsibility of
Subcommittee E01.22 on Statistics and Quality Control.
2
Current edition approved Oct. 31, 1986. Published December 1986. Originally Annual Book of ASTM Standards, Vol 03.05.
3
published as E 158 – 60 T. Last previous edition E 158 – 83. Annual Book of ASTM Standards, Vol 03.06.

Copyright © ASTM, 100 Barr Harbor Drive, West Conshohocken, PA 19428-2959, United States.

1
 E 158
standard procedure still produces better results than the use of
absolute intensities. Accordingly, the method is employed
almost universally.

6. Plotting Analytical Curves with a Constant Internal

Standard
6.1 Log-Log, Linear, and Log-Linear Plots—Different plot-
FIG. 1 The Advantage of Using an Internal Standard ting methods produce analytical curves of widely different
appearance. Each has its own advantages. Fig. 2, Fig. 3, and
Fig. 4 illustrate the curves most commonly employed. The
characteristics of each are listed. It should be emphasized that
a knowledge of the theoretical shape of a curve can assist in
deciding how best to fit a curve to an array of plotted points.

FIG. 2 Log-Log Plot

with spectra of normal complexity, provided that a spec-

trograph or spectrometer with adequate dispersion and resolu-
tion is used, and that analytical lines are selected carefully. In
general, the intensity of an analytical line is a function only of
the concentration of the element that produces it. Therefore, it FIG. 3 Linear Plot
is common practice to employ an individual analytical curve
for each element in a matrix. Observed intensities are con- 6.2 Theoretical Data—The theoretical data used to plot the
verted into concentrations directly from this curve, without any curves are presented in Table 1. It is assumed that throughout
correction for other elements. However, at times, it may be the series of measurements the ratio of the intensity of the
desirable to apply corrections for background or for interfer- background to the intensity of the internal standard line
ences from other elements, as discussed later in this practice. remains constant, at 0.030. At low concentrations, the ratio of
5.2 Internal Standard Method—In the early days of quan- the intensity of the analytical line to the intensity of the internal
titative optical emission spectrochemical analysis, Gerlach and standard line (Ia/Is) is assumed to vary in direct proportion to
Schweitzer4 found that when they measured the “blacknesses” the concentration. At 0.01 %, (Column 2), it is assumed to
of lines produced on photographic emulsion by a series of measure 0.100 (Column 3); therefore at 0.0032 %, it will
chemically analyzed materials and plotted them against their measure 0.032, etc. The total signal is the sum of the intensity
known concentrations, quite erratic results were obtained. A ratios of the background and line. Thus, at 0.01 %, it is 0.130;
single specimen gave different “blacknesses” when run repeat- at 0.0032 %, it is 0.062, etc. The table first lists concentrations
edly. These variations were ascribed to erratic excitation and to down to 0.0001 %, where the line signal is 0.001, and the
variations in illumination. If they measured a line of an background, 0.030. Then the table lists increasing concentra-
“internal standard element” that had the same concentration in tions. At 0.32 %, it is assumed that self-absorption is signifi-
all their materials, they could measure these exposure varia- cant, so that the line response is not as great as the increase in
tions. Ratios determined by dividing the “blackness” of an concentration. The data tabulated in Column 5 are explained
analytical line by the “blackness” of an internal standard line, later in 6.5.1.
gave more repeatable results when the same material was 6.3 Characteristics of Log-Log Plots—Fig. 2 shows the
burned repeatedly. Plots of “blackness” ratios versus concen- appearance and major features of a log-log plot. The intensity
tration produced analytical curves with a better fit of points. ratios of the total signal (Column 4, Table 1) are plotted against
Fig. 1 illustrates the advantages in precision and fit of points to concentrations (Column 2), using log-log graph paper. In the
a curve obtained by the internal standard technique, when illustration, intensity ratios are plotted vertically, since these
applied to the determination of nickel in low-alloy steels. are measured, and concentrations horizontally, since they are
Curves plotted on linear graph paper are used for illustration. known (at the time of plotting the curve). However, some
5.2.1 Over the years, power supplies for sampling and workers prefer to reverse the scales.
exciting materials have become much more stable, and expo- 6.3.1 Below the detection limit, the curve starts as a
sure variations have been reduced, but in general, the internal horizontal line, corresponding to the spectral background. The
curve rises slowly as the line signal can be distinguished above
the background. At approximately 0.01 %, the curve becomes
4
Gerlach, W., and Schweitzer, E., Foundations and Methods of Chemical a straight line at, or close to, a slope of 45°. In other words,
Analysis by the Emission Spectrum, Adam Hilger, Ltd., London, 1929. when the background becomes a small part of the signal,

2
 E 158
trograph or a direct reading instrument is performing as theory
would predict.
6.3.4 Analytically, a log-log plot provides an open scale at
low concentrations, and equal readability at 0.0001 % or 1.0 %.
This is valuable when the concentration range exceeds two
orders of magnitude. Log-log plots are most useful when the
ability to read low concentrations is relatively more important
than accurate readings at high concentrations. An example
would be the analysis of geological samples with a d-c arc,
where an element may vary from 0.001 % or less, to 1.0 % or
more, but the relative precision will never be better than 5 % of
the amount present. Note, however, that a log-log plot is
compressed at high concentrations, such as in measuring
carbon in cast iron in the 2 to 4 % range, where the readability
FIG. 4 Log-Linear Plot
should be 0.01 % or better.
6.3.5 Instrumental variables, such as dirty optics, may
TABLE 1 Theoretical Data for Log-Log, Linear, and Log-Linear introduce a proportional reduction in the intensity. For ex-
Plots ample, there may be a 10 % loss at all intensity levels.
1 2 3 4 5 Cleaning the condensing lens or window may restore the
Intensity original response. Such changes are compensated for by
Ratio, Line “standardizing” the instrument. On a log-log plot, such changes
Intensity Time (t)
Std. + Back-
No.
% Ratio, Line
ground
+ 3000 produce an upward or downward parallel shift of the curve. In
Alone (See 6.5.1)
(Col. 3 + a direct-reading spectrometer, the original response can be
0.030) restored by a “gain” adjustment.
1 0.01 0.100 0.130 5601 6.3.6 A change in the background level will result in a
2 0.0032 0.032 0.062 6545
3 0.0010 0.010 0.040 7104 rotation of the curve and a change in its shape. There will be
4 0.00032 0.0032 0.0332 7342 almost no effect at high concentrations, but a very significant
5 0.00010 0.0010 0.031 7429 change at low concentrations.
6 0.032 0.316 0.346 4353
7 0.10 1.000 1.030 2962 6.4 Characteristics of a Linear Plot—Fig. 3 shows the
8 0.32 2.800 2.830 1674 major features of a linear plot. The intensity ratios are plotted
9 1.00 6.300 6.330 647
10 3.12 8.900 8.930 208
against concentrations using linear graph paper. The solid line
Supplementary Points (See Fig. 6) shows the complete plot, using the left-hand and bottom scales.
11 0.008 0.080 0.110 The low-concentration portion of the curve (dashed line) is
12 0.0065 0.065 0.095
13 0.005 0.050 0.080 plotted on an enlarged scale using the top and right-hand
14 0.0024 0.024 0.054 scales.
15 0.0015 0.015 0.045
6.4.1 The linear graph gives a very open plot at high
concentrations. However, it cannot be read with desired preci-
doubling the concentration approximately doubles the intensity sion at low concentrations (below about 5 % of full scale).
ratio. In a spark or d-c arc, the curve develops a “shoulder” as Larger graph paper helps, but even with 50-cm paper it is not
a result of self-absorption, at 800 to 3000 times the detection possible to read to two-figure precision below 1 % of full scale.
limit, depending on the individual spectral line. Hollow cath- Wider concentration ranges should be plotted in segments, with
odes, glow discharges, inductively coupled plasmas, or similar expanded scales for the lower concentrations, as shown in the
sources may continue a straight line response over a much illustration.
greater range. 6.4.2 At low concentrations, the plot is a straight line,
6.3.2 For a spark or an arc, because of the “toe” and starting at the background reading for zero concentration. At
“shoulder,” three-cycle log graph paper is usually adequate to high concentrations, the self-absorption becomes apparent
record the intensity range even though four or five cycles may more quickly than in a log-log plot, and the curve flattens more
be needed to record the concentration range. dramatically. The straight-line response makes a linear plot
6.3.3 The scales of log-log graph paper are preselected. particularly appropriate to fitting a curve through points that
Photographically, if the intensity ratios have been determined tend to scatter at low concentrations, where the precision
properly, as specified in Practice E 116, the shape and slope of becomes poor. Also, if there are points for intermediate
a log-log plot of intensity ratios versus concentration are not concentrations that fall significantly above the straight line
affected by the contrast of the photographic emulsion or the established by the lowest and one of the higher concentrations,
conditions of development. When using a direct-reading spec- this is a clear indication of probable interference from another
trometer, the voltage applied to the dynodes of the photomul- element. The distance above the line is a measure of the
tiplier, or the gain of an amplifier, will determine the back- amount of interference.
ground level, but will not affect the shape or slope of a log-log 6.4.3 Instrumental changes such as dirty or cleaned optics,
plot. Accordingly, a log-log plot can show whether a spec- which result in proportional loss or gain in intensity, produce a

3
 E 158
rotation of the curve. The rotation is about the intercept with 7. Plotting Analytical Curves with a Variable Internal
the zero intensity ratio axis, which is shown by the dashed line Standard
of Fig. 3. On a direct-reading spectrometer, a “gain” change 7.1 Internal Standard Dilution—In high alloys, where two
can compensate, and restore the curve to its original slope. or more elements vary independently over rather wide concen-
6.4.4 On linear plots, background changes produce vertical tration ranges, the internal standard will no longer have a
curve shifts. constant concentration. For example, in stainless steels, where
6.5 Characteristics of Log-Linear Plots—Log-linear plots nickel and chromium vary widely, the iron may have a
are employed with the type of readout system that has a concentration range from 65 to 75 %. The iron intensity will
logarithmic response. The commercial version utilizes the time vary with the iron concentration as well as with the changes in
that is required by the internal standard capacitor, or its illumination, which it is supposed to measure. Because of this
equivalent, to fall to the voltage accumulated on an analytical “internal standard dilution,” a specimen with 65 % iron will
capacitor. The logarithmic time decay is represented by the produce a low-intensity reading for iron, which will give a
equation: high-intensity ratio, INi/IFe. The point will fall above the curve.
Ei 5 E0e2t/k (1) Specimens with unusually high iron concentrations will pro-
duce points that fall below the curve.
where:
Ei = the voltage accumulated on the analytical capacitor,
E0 = the voltage on the internal standard capacitor,
t = the time required for E0 to decay to Ei, and
k = the RC constant determined by the product of the
capacitance and the bleeder resistance of the circuit.
Long times, or high digital counts, represent low concentra-
tions; short times, or low counts, correspond to high concen-
trations. Thus, the circuitry produces a negative slope. The
theoretical slope is the negative reciprocal of the RC constant.
6.5.1 If Eq 1 is converted to natural logarithms and solved
for t, then,

t 5 2k 3 ln ES D
Ei
0
(2)

In the later commercial instruments with digital readouts, the FIG. 5 A Typical Concentration Ratio Plot
RC constant, k, is adjusted to 1275 counts. In the fifth column
of Table 1, (Ei/E0) represents the ratio of the intensity of the
signal plus background to the intensity of the internal standard. 7.2 Concentration Ratio Plots—To compensate for internal
To maintain the readings positive, a bias of 3000 counts has standard variations, intensity ratios are plotted against concen-
been added. Therefore, Column 5 represents the time, ex- tration ratios.5 To continue the illustration of stainless steels, if
pressed as digital counts: we plot INi/IFe against %Ni/%Fe, the points fall on a smooth
t 5 1275 @–ln ~Column 4!# 1 3000 (3) curve, as shown by Fig. 5. Linear graph paper is used to obtain
an open scale at high concentrations.
For example, the first entry is: 7.2.1 Concentration ratio plotting does place the points on a
1275@2ln 0.130# 1 3000, or 1275@2~22.04!# 1 3000 5 5601 smooth analytical curve, but there is no longer a direct
6.5.2 It is customary to plot concentrations vertically on a relationship between intensity ratios and concentrations for the
logarithmic scale and time or digital counts horizontally. Fig. 4 elements being measured. The concentration of the internal
shows a log-linear plot, using the data from Table 1. standard must be determined first, before the concentration
6.5.3 The log-linear plot, like the log-log, has a theoretical ratios can be converted into concentrations. The concentration
response that is independent of instrumental variables so long of the internal standard can be determined by a formula that is
as the RC constant is maintained the same. Therefore, it shows based on the fundamental rule that the concentrations measured
whether an analytical line is performing as expected. It has a for all significant elements must add up to 100 %. This is
“toe” at the background, changing to a straight line at higher illustrated below for stainless steels, where the element sym-
concentrations. A “shoulder” develops if the line starts to show bols represent their concentrations:
self-absorption. C 1 Mn 1 S 1 P 1 Si 1 Cr 1 Ni (4)
6.5.4 The log-linear response provides an open scale at low 1 Mo 1 Nb 1 T 1 Fe 5 100 %
concentrations, and equal readability at all levels. The time
scale is compressed at high concentrations. where:
6.5.5 Instrumental changes that result in a proportional T = the total trace elements not measured.
change in intensity produce a lateral shift of the analytical
curve.
6.5.6 A change in background alters the shape of the curve 5
Coulliette, J. H., “Spectrographic Determination of Nickel and Chromium in
at the low-concentration end of a log-linear plot. Stainless Steel,” Industrial and Engineering Chemistry, Vol 15, 1943, p. 732.

4
 E 158
For example, it might be assumed that copper, cobalt, 8.1.3 If the background varies significantly from exposure
vanadium, and all other elements not measured would have a to exposure, independently of the line signal or overall expo-
nominal value of T = 0.15 % 6 0.05 %. Dividing Eq 4 by the sure, background correction can improve the precision and
concentration of iron (Fe), gives: accuracy of the analysis and thus can improve the detection
C Mn S P Si Cr Ni limit.
Fe 1 Fe 1 Fe 1 Fe 1 Fe 1 Fe 1 Fe (5) 8.1.4 In a direct reader, background correction may make it
Mo Nb T Fe 100 possible to read concentrations directly, as explained later.
1 Fe 1 Fe 1 Fe 1 Fe 5 Fe 8.1.5 These benefits become real only if the instrument or
the analytical procedure, or both, can produce good long-term
This can be represented in generalized terms by:
stability.
Ci 100 2 T
8.2 Photographic Background Correction—On a photo-
(
Cs 1 1 5 C s (6)
graphic plate or film, it is easy to measure the background
where: beside a line and to ensure that this measurement does not
Ci = the concentration of the individual analytical include lines of possible interfering elements. The approved
elements, method of background correction recognizes and compensates
Cs = the concentration of the internal standard, for the emulsion’s logarithmic response. This involves a rather
Ci = the sum of all the concentration ratios deter- complicated and lengthy procedure which is described in detail
(
Cs mined determined from the individual analyti- in Practice E 116. Such corrections have been used to improve
cal curves, and, as before, and the detection limits of trace elements in uranium.
T = the concentration assumed for the total of all the 8.3 Background Correction with Direct Readers—At low
trace elements not measured. concentrations, a direct reader gives an overall reading that is
The concentration of the internal standard can be calculated largely background and partly signal, but there is no simple
from Eq 6 as: means of distinguishing which is which. On the other hand,
100 2 T because of the linear response of photomultipliers, at low
Cs 5 Ci (7) concentrations both the line signal and the background respond
11(C
s linearly, so that background correction can be done more
7.2.2 The concentration ratio method requires that all ele- simply than photographically. It consists of a simple subtrac-
ments be measured if they are likely to occur in concentrations tion. Three methods of background correction are employed:
above 0.3 %. If the total of the trace elements not measured is 8.3.1 With static background correction, a fixed number is
not likely to exceed 0.5 % and if the total is not expected to subtracted from all readings. For example, for the data of Table
vary by more than 0.1 %, T can either be represented by the 1, 0.025 might be subtracted from all of the readings of
probable total as a constant, or the term can be disregarded, Column 4. If negative readings are considered objectionable,
without affecting any measurement significantly except the subtract slightly less than the full background reading of 0.030,
concentration of the internal standard. because otherwise some negative readings will be obtained as
7.2.3 Once the concentration for the internal standard has a result of statistical fluctuations in the data. Alternatively, if a
been calculated, the other concentrations can be computed by: computer is being used, the full background correction can be

S D
applied, and the software can be written either to instruct the
Ci
Ci 5 C Cs (8) computer to print a zero instead of a negative intensity ratio, or
s
to indicate that the reading falls below the lowest meaningful
NOTE 1—Coulliette originally included the concentration ratio method intensity ratio.
of calculation as Section 9 in previous editions of Practice E 158, but it
was never used extensively either in photographic or direct-reading
8.3.2 With dynamic background correction, the background
analysis because reading the concentration ratios from the curves, and is measured in each exposure by one or more exit slits, which
making the calculations added substantially to the time of analysis. are profiled to make sure that they show no line response to any
However, with the introduction of programmable calculators or comput- of the elements in the matrix. A portion of the background
ers, the calculations became practicable. reading is subtracted from an analytical channel to obtain a net
figure for the line itself. For example, a background slit might
8. Background Correction read 0.150. For the data in Table 1, a potentiometer or other
8.1 Utility—Background correction is appropriate only control would be adjusted to deduct 0.025 to 0.029 from the
when working close to the detection limit. Then it offers the analytical channel. The full correction of 0.030 can be avoided
following advantages: if desired, to prevent frequent negative readings. With a
8.1.1 On a log-log plot, background correction will extend computer readout, a correction factor for the background can
the straight-line response to lower concentrations. The appar- be selected to apply a similar correction. Also with a computer
ent improved response provides better readability. If the same readout, negative numbers can be avoided as in 8.3.1.
data are plotted linearly there will be no benefit; the curve will 8.3.2.1 If the specimens being analyzed will always have
merely be moved down toward the origin. the same matrix, a single-background channel will suffice for
8.1.2 If the background under an analytical line increases dynamic background correction at all wavelengths. A typical
proportionally to a matrix element, background correction can example is the determination of trace impurities in uranium
improve the accuracy of analysis. oxides.

5
 E 158
8.3.2.2 When the major elements vary over wide concen- tration if divided by 10 000.
trations from specimen to specimen, the background measured
at one wavelength may not reflect the background variation at 9. Concentration Calculations
a distant wavelength, because each of the major elements has 9.1 Conversion of Intensity Ratios into Concentrations—
its own distribution of background with wavelength. Therefore, The first step in analyzing an unknown sample spectrochemi-
in making dynamic background corrections for specimens with cally is to determine the ratios of the intensities of the
a highly variable matrix, the background channels should be analytical lines to the intensities of the internal standard line
within 300 to 500 Å of the lines they are correcting. An
example would be the analysis of trace elements in crushed
rock samples, using a d-c arc, where there may be tenfold or
S D
Ia
Is
greater changes in such matrix elements as silicon, calcium, whether this is done photographically or by a direct-reading
magnesium, aluminum, iron, sodium, and potassium. The spectrometer with an analog or digital readout system. Then
situation may be complicated further with a d-c arc if different these intensity ratios can be converted into concentrations by
exposure times are employed for volatile, intermediate, and reading from any of the analytical curves illustrated in Fig. 2,
refractory elements. Then the background channels and the Fig. 3, or Fig. 4. However, this is time-consuming and easily
analytical channels that they correct should have the same results in errors. Therefore, it is common practice to prepare
exposure conditions. There are no definite guidelines for concentration tables, or scales for a calculating board, from
background correction; consequently optimum conditions have which the concentrations may be read directly, with less error.
to be determined empirically for each line. Another arrangement with a direct-reading spectrometer is to
8.3.2.3 With dynamic background correction, if the back- have a strip chart recorder or x-y recorder, in which the pen
ground varies independently of the analytical line, background deflects linearly with intensity ratios. A series of concentration
correction may improve the precision with which the line alone scales preprinted on the paper makes it possible to read the
can be measured. This should improve the repeatability of concentration of each element.
multiple readings for a specimen and the fit of points to a 9.1.1 Concentration tables are most appropriate for short
smooth curve at low concentrations. It may also improve the concentration ranges and for a limited number of matrices. A
detection limit. However, if the errors in measuring the single flip card can list intensity ratios and the corresponding
background and the overall signal are greater than the variation concentrations for each element.
of the background independently of the signal, the detection 9.1.2 Concentration scales for calculating boards or pre-
limit will be poorer. printed paper are readily prepared from analytical curves. Fig.
8.3.3 Some device can be employed to shift automatically 6 and Table 2 illustrate the process.
from the line plus background to the background alone, many 9.1.2.1 Usually the analyzed specimens do not have appro-
times during an exposure. Then the integrated background can priate concentrations for preparing a concentration scale. The
be subtracted from the overall signal for the line plus back- concentrations may be 0.175, 0.042, 0.073, etc., and for a scale
ground, to produce a reading for the line alone. Since it is initially we want major divisions: 0.010, 0.020, 0.030, 0.040,
difficult to provide different shifts for each analytical line, some 0.050, etc.
compromise distance has to be selected, and this may result in 9.1.2.2 Fig. 6 illustrates how an analytical curve can be used
a line interference for some other element. Background cor- to prepare a concentration scale. It shows an enlarged section
rection should not be applied to such a channel. of the log-log theoretical curve, Fig. 2, from 0.0010 to
8.3.3.1 As in 8.3.2.3, this background correction may or 0.010 %. The original points, as circles, have been supple-
may not improve precision or the detection limit, or both. It can mented by the points shown as squares. Column 1 of Table 2
improve accuracy, as interference correction does. lists concentrations appropriate to constructing a concentration
8.3.4 With a direct-reading spectrometer with linear re- scale. Column 2 lists the corresponding intensity ratios read
sponse, background correction can be employed to produce a from the curve.
direct concentration readout. First, background correction is 9.1.2.3 To make the concentration scale, proceed as follows:
applied so that the reading for a blank specimen starts at or near Mount a blank for the scale, AB, at the right-hand side of the
the origin. If the response is linear, with no self absorption, a plot parallel to the intensity ratio scale. (This will produce a
“gain” adjustment can be applied to a high-concentration scale that will read from low to high concentrations from left
specimen until the digital reading matches the concentration. to right.) Mark a line opposite each intensity ratio that
Then all intermediate specimens should read directly in con- corresponds to one of the preselected concentrations, as in
centration. For example, with a digital meter reading from 1 to Table 2. Use a longer line (3 mm in the illustration) for each
10 000, assume that a “blank” reference material for copper in major concentration: 0.001, 0.002 %, etc. Use an intermediate
zinc die cast alloys reads 250. The blank is known to have a line (2 mm) for each fifth division: 0.015, 0.025, 0.035, etc.
copper concentration of 0.0006 %, and the copper analytical Use a short line (1 mm) for the units: 0.012, 0.017, 0.022,
curve is a straight line on linear graph paper up to 0.3 %. 0.027, etc. Insert the remaining units by eye, that is, 0.011,
Background correction can be applied to bring the reading for 0.013, 0.014, 0.016, etc. This process is illustrated only for the
the blank to 6. Then the gain adjustment can be used to make range from 0.011 to 0.019. It needs to be continued the whole
the reading for a reference material with 0.24 % copper read length of the scale. Number the major divisions, as illustrated.
2400. All intermediate specimens will read directly in concen- Identify the scale. In the illustration, nickel has been used.

6
 E 158

FIG. 6 Preparation of a Concentration Scale

TABLE 2 Concentration and Intensity Ratio Readings for a

Concentration Scale
Concentration, % Intensity Ratio
0.0010 0.040
0.0012 0.042
0.0015 0.045
0.0017 0.047
0.0020 0.050
0.0022 0.052
0.0025 0.055
0.0027 0.057
0.0030 0.060
0.0032 0.062
0.0035 0.065
0.0040 0.070
0.0045 0.075
0.0050 0.080
0.0055 0.085
0.0060 0.090 FIG. 7 Representation of a Curve by Straight-Line Segments
0.0065 0.095
0.0070 0.100
0.0075 0.105 orders have been used most frequently to represent analytical
0.0080 0.110 curves. For example, the concentration can be represented by a
0.0085 0.115
0.0090 0.120 fourth-order polynomial:

S D S D S D S D
0.0095 0.125 Ia Ia 2 Ia 3 Ia 4
0.010 0.130 C5a1b I 1c I 1d I 1e I (9)
s s s s

If the coefficients c, d, and e are made zero, the equation

9.2 Concentration Calculations with Computers or Pro- reduces to a straight line on linear graph paper:
grammable Calculators—With computers or programmable
calculators, the conversion of intensity ratios into concentra-
Ia
C5a1b I
s
S D (10)
tions can be made automatically. Each analytical curve must be
stored in memory. This means that the curve must be repre- where:
a = a negative number representing the intercept of the
sented either by a series of straight-line segments or by some
curve with the concentration axis. In Fig. 3 (dashed
arithmetic formula.
curve) it is − 0.003 %.
9.2.1 The curved section of the analytical curve can be
b = the tangent of the angle that the straight line makes with
represented by a series of straight lines, as illustrated in Fig. 7.
the intensity ratio axis.
The straight lines are stored in memory by the coordinates of
In Fig. 3, this tangent can be calculated from the coordinates
their intersections, A, B, C, D, and E. The measured intensity
at C = 0.10 and 0 % as
ratio of an unknown is converted by the computer into
concentration by a simple linear interpolation along the appro- 0.100 2 0
priate line segment. 1.00 2 0.03 5 0.1031
9.2.2 A linear equation or polynomials of second or higher Thus, the equation becomes:

7
 E 158
Ia
C 5 20.003 1 0.103 I
s
S D (11) where:
CT = the theoretical concentration, represented
Note that Eq 10 represents the theoretical response at low by extending the straight line response
concentrations, and so is particularly appropriate when work- which would be obtained without self ab-
ing in the range from 2 to 500 or 700 times the detection limit sorption, and
for the line. It is appropriate also for higher concentrations, if AR and MR = two constants that must be calculated by
the concentration range is sufficiently short to permit the curve simultaneous equations from at least two
to be represented by a straight line, within the precision of the well-separated points in the reversal (self-
measurements. The segment B–C in Fig. 7 is an example. absorbed) section of the curve.
9.2.2.1 If a single curve is required to cover a concentration CT may be calculated from the formula:
range greater than a straight line can represent within the CT 5 10mBt 1 AB – G (13)
experimental error, the c, d, and e coefficients can be utilized,
in that order. A value for c can be calculated that will extend the where:
curve over a short additional concentration range, or over a mBand AB = two constants to be evaluated by simulta-
short range at higher concentrations that is definitely a curve. neous equations from at least two well-
With appropriate software, a computer can calculate the separated points in the lower concentration
coefficients. straight-line response region (which includes
9.2.2.2 When the d and e coefficients are utilized, it is the extended background-corrected section),
common practice to ask the computer to calculate them, with a t = the time of logarithmic decay (the intensity
least squares or weighted least squares fit. Unless limitations ratio in the logarithmic system), and
are placed on the values of the coefficients, the computer may G = background expressed in %.
produce a curve that weaves in and out between the experi- 9.2.3.1 With a linear readout, Kennedy’s equations become:
mental points; also, it may double back on itself at low and C 5 CT 1 kCTm (14)
high concentrations. This can produce erroneous results unless
the operator ensures that the low and high intensity ratios at where:
which this occurs are beyond the range of any possible k = a coefficient,
measurements. Finally, with a standard least squares fit, a m = an exponent, both to be evaluated by simultaneous
computer will try to match the curve to the points at 1000 or equations from two well-separated points in the
2000 ppm as closely as it matches points at 1 or 2 ppm. self-absorbed region, and
Analytically, it may be important to read fractional ppm at the CT = the theoretical straight-line response, given by the
low range, whereas 10 or 20 ppm is all that is significant at formula:
high concentrations. One way to offset this tendency is to
employ heavier weighting for the low points or to use many
Ia
CT 5 a 1 b I S D
s
(15)
more standards at the low concentrations than at the high. as discussed in 9.2.2. The general validity of Kennedy’s
9.2.2.3 There are several ways of forcing the computer to equations has been confirmed by a recent study of the response
conform more closely to the theoretical curve. For example, the of several common analytical lines far beyond the range for
computer can be instructed that a should be negative, and that which they are considered appropriate—nickel 2316 Å to 12 %
b, c, d, and e should all be positive. Then the computer cannot and chromium 2677 Å to 20 %, in ferrous matrices; and silicon
produce a curve that weaves or reverses. Or the software can be 2516 Å to 12 % in aluminum alloys. Typical performance for
written to report any undulations in the curve. Then the the chromium 2677Å line is shown by Fig. 8. The normal
operator can explore the seriousness of any error that might be straight-line section will have a slope of almost 45°. Then, as
introduced and can employ a lower degree equation or some self-absorption becomes significant, a curve region occurs.7 At
other method to improve the curve fit. Curves so restricted will still higher concentrations, the rate of self-absorption tends to
match more closely the response of sensitive lines. become constant and the plot continues as approximately a
9.2.3 A model equation that matches the actual response of straight line with a reduced response, with a slope of 30° or
analytical lines is better than a polynomial, which requires an less.
artificial fit. Kennedy6 has studied this, and points out that if
dynamic background correction is applied to both the analyti- NOTE 2—These equations do not apply to truly self-reversed lines, such
as lithium 6707 Å, potassium 7665 Å, sodium 5890 Å, magnesium 2852Å
cal line and the internal standard, the curve will start at the , cadmium 2288 Å, or zinc 2138 Å, excited in a d-c arc and measured with
origin. At higher concentrations, the actual concentrations are a narrow exit slit. Such lines show a real reversal of response. Then an
always greater than the theoretical straight-line response, due intensity ratio near the top reading becomes ambiguous; it can indicate a
to self-absorption. The self-absorption increases exponentially concentration at the low side of the peak or a much higher concentration.
with the concentration. For a logarithmic readout system, When such a self-reversed line is employed to analyze specimens that may
Kennedy derived the equation: have concentrations in the self-reversed range, its use should be termi-
nated at a concentration low enough to avoid ambiguity. Another less
C 5 CT 1 10ARCTMR (12) sensitive line, exhibiting little or no self absorption, should be used to

6 7
Kennedy, W. R., “Where’s My Curve?,” Computers in Analytical Chemistry, Cowan, R. O., and Dieke, G. H., “Self-Absorption of Spectrum Lines,” Review
Plenum Press, New York, NY, 1969. of Modern Physics, Vol 20, 1948, p. 418.

8
 E 158
measure the higher concentration. 9.2.5.1 It is more practical to develop a computer program
so that the computer can calculate the constants, with some
form of least squares fit to the individual points, and it is
desirable to have the computer report the one or two points that
give the poorest fit, so that these can be double-checked. If a
specific reason can be found for considering them non-
representative, they can be discarded and the constants can be
recalculated. The program should make it possible to enter test
intensity ratios and obtain calculated concentrations, so that the
curve can be plotted. Even better, the computer can be
programmed to display the plot of the measured points and the
curve that it has calculated, so that the operator can tell at a
glance whether the solution is a good one.
9.2.5.2 It is desirable to terminate curves at the low and high
FIG. 8 Typical Response of a Sensitive Line over a Wide concentrations at which they are considered acceptably accu-
Concentration Range rate, and to have the computer print < the low limit and > the
higher limit.
9.2.4 In most empirical work aimed at finding a formula to 9.3 Use of Multiple Lines—As indicated by Fig. 8, at low
represent a typical analytical curve, concentration has been concentrations the intensity ratio almost doubles when the
treated as being a function of intensity ratio. Crawford and concentration doubles. At the high end of the curve it does not
Beaty8 say that theoretically this should be reversed because, in quite double when the concentration quadruples. Therefore, no
the analytical gap, the intensity is determined by concentration. matter how closely the curve fits the points at the higher
On this basis, they derived a modification of the exponents of concentrations, the standard deviation in concentration will be
a second degree equation. Their proposed equation can be more than double the standard deviation in intensity ratio.
restated in a somewhat simpler form as: Obviously, well before the response has fallen to a slope of 30°,
it is better to shift to a less sensitive line, with a higher
Ia/Is 5 a0 1 a1Cn 1 a2C2n (16)
background but a 45° slope. If the complete concentration
where: range from the detection limit to 100 % is required for a
a0 = the background intercept with the C = 0 axis, particular element, two or three lines are needed to obtain
a1 = the slope at low concentration, which has to be optimum accuracy. Examples might be copper or nickel, which
evaluated, must be determined at the lowest possible concentrations in
a2 = a constant, which is always negative, and is evaluated zinc die cast alloys or in high-purity lead, at intermediate
to define the curvature, and ranges in aluminum alloys, and then must serve as internal
n = an exponent, normally in the range from 0.90 to 1.0, standards in copper- or nickel-base alloys. However, some-
which has to be evaluated. times multiple lines are not possible, as in the case of carbon in
This equation was found to fit empirical points for standard steels and cast irons, where only the carbon 1931 Å line gives
reference materials, such as those plotted in Fig. 8, to better adequate response in the available wavelength range. Unfortu-
than 1 %, well within the experimental error. nately, the carbon response begins to curve in the 2 to 4 %
9.2.4.1 Concentration can be calculated directly by invert- range, just where it is important in cast iron. Therefore in
ing Eq 16 to: measuring high-carbon levels in cast iron, the spectrochemical

C 5 S 2a1 1 =a1 2 2 4a2@a0 2 ~Ia/Is!#

2a2 D 1/n

(17)
parameters should be adjusted to minimize self-absorption.
10. Corrections For Interferences From Other Elements
9.2.5 Eq 9, Eq 10, Eq 12, Eq 14, or Eq 16 can be fitted to an 10.1 Causes of Interference from Other Elements—In opti-
array of plotted points by calculating the required constants. In cal emission spectrochemical analysis, usually interferences
trial-and-error fitting, this is simple only for a straight line. For from other elements are significant only when making mea-
curved lines, constants must be assigned estimated values, surements at low concentrations, near the detection limit. At
starting from the straight line. Specific points should be high concentrations, there are many analytical lines that can be
calculated and the corresponding curve should be plotted. used for most elements. Generally, a spectral line can be
Usually the first trial will not fit the points well at some section selected that will give good response and precision, and be free
of the curve. New values for the constants should be assigned of interferences from elements likely to be encountered at
and the process repeated until the curve fits the points to the significant levels in any of the material to be analyzed.
experimental error. The procedure is lengthy and tedious, and Interferences are caused by:
becomes a major task when there are many analytical curves to 10.1.1 A line of the interfering element that is not resolved
calculate. by the working resolution of the spectrometer.
10.1.2 A change in the background under the analytical line,
8 due to the interfering element.
Crawford, R., and Beaty, J., “An Advanced Curve Fitting Technique for a
Direct Reading Spectrometer,” Pittsburgh Conference on Analytical Chemistry and 10.1.3 An enhancement or suppression of the analytical line
Applied Spectroscopy, 1978. by some physical interaction in the analytical gap between the

9
 E 158
interfering element and the element being analyzed. For 10.3.3 If the procedures above do not reveal strong suspects,
example, sulfur above 0.1 % in steels may produce inaccurate a systematic search can be conducted. For each element that is
readings for itself and for manganese unless the pre-burn and likely to occur at a fairly high concentration in some of the
exposure times are properly selected. Carbon above 0.1 % may materials to be analyzed, prepare a mixture with graphite
affect sulfur also. Alkali metals in a d-c arc will drop the powder (or a solution) that has that element at the highest and
temperature of the discharge, which will reduce the intensity of lowest concentrations at which it is likely to occur. These
analytical lines in the far ultraviolet, such as zinc at 2138 Å or mixtures should be burned, and the readings on each analytical
cadmium at 2288 Å. There are other instances of such channel should be compared. When the high concentration
interference that are not so well understood. Evidence of this produces a significantly higher reading than the low, a probable
enhancement or suppression is a change in slope of the interference is indicated.
analytical curve at different concentrations of the interfering 10.3.3.1 For example, in the analysis of silicate rocks,
element. calcium, magnesium, aluminum, iron, silicon, potassium, and
10.1.4 An interference may affect the internal standard line. sodium will be major elements. Titanium, manganese, and
When this occurs, a specimen with an unusually high concen- phosphorus will be minor elements and all other elements will
tration of the interfering element will have an unusually high be traces. The major and minor elements should be investi-
reading for the internal standard line. This will produce gated as potential contributors of interference, as follows:
unusually low intensity ratio readings for all major elements. 10.3.3.2 Iron oxide is not likely to occur above 20 % nor
Thus, if one or several reference materials seem to produce below 1 % in silicates. Pure silicon oxide can be used as a base
points below the smooth curves established by the other and two mixtures can be prepared, with 30 % and 1 % iron
reference materials, for all major elements, it is an indication of oxide. These can be buffered with graphite, sodium carbonate,
possible interference on the internal standard line. and graphite, or whatever other material will be used to buffer
NOTE 3—Interference on the internal standard line should not be the analytical specimens. The two mixtures can be burned. As
confused with internal standard dilution, where one or several samples an example, if the 30 % mixture produces a higher reading on
may have abnormally high concentrations of the internal standard. It is the channel for the 2598.06 Å antimony line than the 1 %
assumed that internal standard dilution effects will have been compensated mixture, there is an indication of serious interference from the
for by concentration ratio plots, as described in Section 7. 2598.37 Å iron line. This can be confirmed by profiling.
10.1.5 Molecular band spectra can introduce interferences. 10.3.3.3 The procedure is repeated for each of the other
Notable examples are CN bands in d-c arc or spark spectra in major and minor elements, using concentrations in the graphite
air, or the OH bands in spark or plasma spectra of solutions. that match the highest concentrations likely to be encountered
10.2 Reason to Suspect Other-Element Interferences— in the specimens to be analyzed. Quite a few suspected
When an analytical curve is being prepared, if there are some interferences may be revealed.
points that fall significantly above (or below) a smooth curve 10.3.3.4 The procedure can be repeated for the trace ele-
through the majority of the points, there is a suggestion of ments, again using concentrations that are likely to be encoun-
other-element interference. It then becomes necessary to con- tered in the specimens. A few additional interferences may be
firm that there is, in fact, an interference. revealed.
10.3 Identifying Interfering Elements—Following are sev-
NOTE 4—In looking for interferences in this way, the synthesized
eral methods for identifying interfering elements: specimen containing the suspected interfering element must be free of the
10.3.1 Examine the analysis of the reference material that element to be analyzed. If iron oxide with 0.01 % manganese were used,
falls farthest off the curve, and note whether it contains some obviously it would produce a reading on the manganese channel, but this
other element at a higher (or lower) concentration than in all would not indicate interference.
other reference materials. Note whether the same element is 10.3.4 To detect the interference described in 10.1.3, ana-
present at unusually high (or low) concentrations in the other lyze four specimens which have the following compositions:
reference materials whose points fall off the curve. If there 10.3.4.1 Low concentrations of both the element to be
seems to be a displacement from the curve that is proportional analyzed (A) and the suspected interfering element (I).
to the element, there is a probability of interference.
10.3.4.2 A low concentration of (A) and a high concentra-
10.3.2 Refer to a comprehensive wavelength table,9 and
tion of (I).
look for elemental lines within two times the working resolu-
10.3.4.3 A high concentration of (A) and a low concentra-
tion of the spectrometer, for elements that could occur in
tion of (I), and
appreciable concentrations in the samples. Unfortunately, no
wavelength table can include all elemental lines, and the 10.3.4.4 High concentrations of both (A) and (I). If the
intensity ratings are not reliable. Therefore, interferences may measured spectral responses obtained with specimens 10.3.4.1
occur from lines not listed, or, lines that are listed may and 10.3.4.2 are nearly equal, but the responses obtained with
introduce no measurable interference in the matrix analyzed, specimens 10.3.4.3 and 10.3.4.4 are different (assuming that
with the excitation conditions used. Background variations are the concentration of the element to be analyzed is approxi-
not listed in tables. mately equal in specimens 10.3.4.3 and 10.3.4.4), a spectral
interference of this type is confirmed. The change in response
is roughly proportional to the product of the concentrations of
9
Harrison, G. R., MIT Wavelength Tables, Technology Press, John Wiley, New both the determined and interfering elements.
York, NY, 1969. 10.4 Confirming a Line Interference—Suspected interfering

10
 E 158
elements can be confirmed by burning pure or high concentra- 10.6.1.2 The lead line lies between two weak iron lines that
tion specimens, as in 10.3.3. If it is a case of suspected line are listed in the MIT Wavelength Tables—iron, 4057.35 Å and
interference, a profile plot should be made to confirm that there 4058.23 Å. The lead peak at 4057.82 Å for BAS 59 shows
is a peak at the appropriate position. strongly above the iron background. The peak is drawn toward
10.5 Interference Due to Changes in Background— the 4058.23 Å iron line by the wing of that line, as can be seen
Background interferences usually are significant only for by the theoretical positions noted at the bottom. Manganese at
elements that vary over a wide concentration range of some- 16 % produces a strong signal, not resolved from the 4058.23
thing like 500- or 1000-fold. Calcium and magnesium are Å iron line, which makes the line appear asymmetrical and
notable examples in natural waters and in geological analyses, displaces the peak from its theoretical position. The manganese
because they will increase the background under analytical signal is so strong that it confirms a serious interference on the
lines in the far ultraviolet region or under lines within 50 to 100 lead line.
Å of their strongest lines. Such interference can be confirmed 10.6.1.3 A trial correction (in concentration, as explained
by synthesizing reference materials for the suspected interfer- later) for manganese only, produces a much better fit of points
ant and determining whether they will produce a response to a straight line. NIST 1264 still falls considerably off the
curve with readings from the analytical channel. curve, and NIST 1262 falls off to a lesser extent, as shown by
10.6 Calculating Interference Corrections Manually—The the crosses in Fig. 9, which are plotted from Columns 3 and 6
procedure for making interference corrections is illustrated by
of Table 4. According to the analysis sheet, NIST 1264 has the
the calculation of the interference corrections for the 4057.82
highest concentrations of carbon, molybdenum, and titanium.
Å lead line in the analysis of low-alloy steels.
The second highest concentration of molybdenum is in NIST
10.6.1 When the 4057.82 Å lead line is measured for the
1262 and the MIT Wavelength Tables list a molybdenum line
NIST standard reference materials, numbers 1261–1265,
at 4057.58Å . A small piece of 99.9 % molybdenum, too small
supplemented by BAS standard reference materials, numbers
for multiple exposures, produces a visual profile with a peak at
56–59, the data tabulated in Columns 1, 2, and 3 of Table 3 and
4057.77 Å, much closer to the lead line, but certainly not lead.
Table 4 are obtained. The open circles in Fig. 9 represent the
NIST data, while the open squares are the BAS points. Since This profile is sketched on Fig. 10. Additional profile studies
the highest concentration is only 0.050 %, a straight-line with NIST 1263 and 1264 make it possible to reconstruct the
response would be expected, so it is immediately apparent that contributions of lead, manganese, and molybdenum to the
there must be some kind of interference taking place. The point observed curves. Thus, both manganese and molybdenum are
for NIST 1263 requires the greatest correction, followed by confirmed as interfering elements.
1262 and 1261 and 1264. According to the analysis sheet for 10.6.2 There are two ways of making interference correc-
the NIST reference materials, 1263 has 1.50 % manganese, tions. If the responses of both the element being determined
1262 has 1.04 %, and 1261 has 0.66 %. In the MIT Wavelength and the interfering element are linear, interference corrections
Tables,8 there is a manganese spark line listed at 4057.95 Å, can be made by subtracting an intensity ratio correction from
only 0.13 Å away from the lead line. The two lines will not be the measured intensity ratio, as illustrated by the arrow A for
resolved, since the working resolution of the spectrometer with NIST 1263 in Fig. 9. Alternatively, at low concentrations,
its 25-µm entrance slit and 50-µm exit slit is about 0.34 Å. where the response is a straight line, a concentration correction
Thus, there is strong evidence of manganese interference. can be added to the stated concentration, to obtain the
10.6.1.1 Fig. 10 shows profile curves obtained on the lead “apparent concentration” that results from the interference.
channel for the following three reference materials: NIST This is shown by arrow B in Fig. 9. For an individual
1265, which has less than 0.0001 % lead and 0.0057 % laboratory, there is little to choose between correcting intensity
manganese; BAS 59, which has 0.050 % lead and only 0.12 % ratios or correcting concentrations. However, from the stand-
manganese; and BAS 491, which has 16.1 % manganese and point of a manufacturer of equipment or of a user with several
no listed lead. equivalent instruments, the concentration correction is more

TABLE 3 Intensity Ratio Corrections of Pb Line, 4057.82 Å, for Interferences by Mn, 4057.95 Å, and Mo, 4057.58 Å, in Low-Alloy Steels
1 2 3 4 5 6 7 8 9
Mn Correction Total
Standard Stated Stated Stated Mo Correction Corrected
Measured kIMn = 0.0079 Correction
Reference Concentration Concentration Concentration kIMn = 0.008 IPb/IFe(Col.
IPb/IFe (0.0079 3 Col. (Col. 5 +
Material of Pb, % of Mn, % of Mo, % (0.008 3 Col. 6) 3 − Col. 8)
4) Col. 7)
NBS
1261 0.000025 0.0449 0.66 0.0052 0.19 0.0015 0.0067 0.0382
1262 0.00043 0.0477 1.04 0.0082 0.068 0.0005 0.0087 0.0390
1263 0.0022 0.0522 1.50 0.0119 0.030 0.0002 0.0121 0.0401
1264 0.024 0.0658 0.255 0.0020 0.49 0.0039 0.0059 0.0599
1265 0.000015 0.0383 0.0057 0.000045 0.005 0.00004 0.0001 0.0382
BAS
56 0.014 0.0533 0.32 0.0025 (0.007) 0.0001 0.0026 0.0507
57 0.010 0.0487 0.16 0.0013 (0.004) ... 0.0013 0.0474
58 0.015 0.0543 0.43 0.0034 (0.005) ... 0.0034 0.0509
59 0.050 0.0821 0.12 0.0009 (0.005) ... 0.0009 0.0812

11
 E 158
TABLE 4 Concentration Corrections of Pb Line, 4057.82 Å, for Interferences by Mn, 4057.95 Å, and Mo, 4057.58 Å, in Low-Alloy Steels
1 2 3 4 5 6 7 8 9
Apparent
Mn Correction Apparent
Standard Stated Stated Stated Mo Correction Concentration
Measured kCMn = 0.0090 Concentration
Reference Concentration Concentration Concentration kCMn = 0.010 Pb, % (Col.
IPb/IFe (0.0090 3 Col. Mn Correction
Materials of Pb, % of Mn, % of Mo,% (0.010 3 Col. 7) 2 + Col.
4) Only
5 + Col. 8)
NBS
1261 0.000025 0.0449 0.66 0.0059 0.0059 0.19 0.0019 0.0079
1262 0.00043 0.0477 1.04 0.0094 0.0098 0.068 0.0007 0.0105
1263 0.0022 0.0522 1.50 0.0135 0.0157 0.030 0.0003 0.0160
1264 0.024 0.0658 0.255 0.0023 0.0263 0.49 0.0049 0.0312
1265 0.000015 0.0383 0.0057 0.0001 0.0001 0.005 0.0001 0.0002
BAS
56 0.014 0.0533 0.32 0.0029 0.0169 (0.0070) 0.0001 0.0170
57 0.010 0.0487 0.16 0.0014 0.0114 (0.0044) ... 0.0114
58 0.015 0.0543 0.43 0.0039 0.0189 (0.0047) ... 0.0189
59 0.050 0.0821 0.12 0.0011 0.0511 (0.0046) ... 0.0511

FIG. 9 Correction of Pb Line, 4057.82 Å for Interferences by Mn, 4057.95 Å, and Mo, 4057.95 Å in Low-Alloy Steels

valuable. Intensity corrections will be different from instru- squares. Note that both methods should produce the same
ment to instrument, depending on the sensitivities of individual corrected curve. Note also how closely the points can be made
photomultipliers and the dynode voltages, which may be to fit a smooth curve, provided that the chemical analyses for
difficult to standardize. Concentration correction factors are all elements involved have the accuracy of these standard
more fundamental, and should be very nearly the same from reference materials.
one instrument to another of the same model, with the same 10.6.3.1 To explain Table 3 more fully, Column 4 lists the
entrance and exit slit widths. Accordingly, once concentration manganese concentrations. Column 5 tabulates the intensity
correction factors have been calculated, they can be applied to ratio corrections obtained by multiplying Column 4 by the
instrument after instrument, with only a slight refinement of the manganese correction factor 0.0079. Column 6 lists the stated
factor. If a concentration correction factor is calculated to be molybdenum concentrations, which are multiplied by the
0.01, this means that 1 % of the interfering element will add molybdenum correction factor 0.008, to obtain Column 7.
0.01 % to the true concentration of the analyte. Columns 5 and 7 are added to obtain the total corrections listed
10.6.3 Table 3 shows the final calculated data for intensity in Column 8, which are subtracted from the measured intensity
corrections applied to the 4057.82 Å lead line. In Fig. 9, the ratios of Column 3, to produce the corrected intensity ratios of
stated concentrations of Column 2 are plotted against the Column 9. The data in Table 4 are similar, except that the
corrected intensity ratios of Column 9, as shown by the X’s. corrections of Columns 5 and 8 are added to the stated
Table 4 shows the final data for concentration corrections. In concentrations of Column 2 to produce the“ apparent concen-
Fig. 9, the measured intensity ratios of Column 3 are plotted trations” of Column 9.
against the “apparent concentrations” of Column 9. The NBS 10.6.3.2 The final data of Table 3 and Table 4 were obtained
points are shown as solid circles and the BAS points as solid only after initial “trial-and-error” calculations, which were

12
 E 158

FIG. 10 Profiles to Confirm Interferences on the Lead Line, 4057.82Å , in Low-Alloy Steels

performed manually. In undertaking such a calculation, some Then, this adopted correction factor is multiplied by the
clues must be sought first that will permit drawing a prelimi- concentrations of the interfering element in all reference
nary curve. In this case, the concentrations of all elements in materials as illustrated by Column 5 of Table 4. In concentra-
NBS 1265 are so low that it can be considered as representing tion correction, these corrections are added to the stated
the low end of the curve. Also, in this instance BAS 59 has the concentrations to obtain the apparent concentrations, as in
highest lead, with low manganese and negligible molybdenum. Column 6 of Table 4, for manganese corrections only.
As mentioned earlier, the lead concentrations are all below 10.6.3.4 When there appears to be more than one interfering
0.1 %, so the curve should be a straight line, starting with NBS
element, as in the example, probably the best approach is to try
1265 and going just to the right or slightly below BAS 59. With
correcting first for the most serious interference. The manga-
only the NBS data, the picture would not be quite so clear. NBS
nese seems to have the greatest effect in the example because
1264 would become the highest lead point, and it has low
manganese but the highest molybdenum. It would be clear that the concentration is over 1 %, whereas the highest molybde-
the curve should be to the right or below NBS 1264, but by num concentration is 0.49 %. In fact, before the final curve was
how much would have to be estimated. obtained, points were plotted just for manganese correction, as
listed in Column 6 of Table 3. This disclosed the need for the
NOTE 5—Data will vary from interference to interference and from one molybdenum correction, as mentioned earlier.
set of reference materials to another, but always there is some indication,
such as that described in 10.6.3.2, which serves as a guide to drawing a 10.6.3.5 After an initial trial curve has been plotted, the
trial curve. points for the standard reference materials with the highest
NOTE 6—If we had only the BAS standard reference materials, it would concentrations should be examined. If they fall to the right of
not be obvious that there was any interference, but a close inspection the best curve through all the points, on a concentration-
would suggest a problem from manganese. The molybdenum interference
would be undetectable.
corrected curve, the correction factor should be reduced, the
calculations made again, and a new curve plotted. If these
10.6.3.3 When there is only one element interfering, and points fall to the left, the factor should be increased. If there are
when only a few standard reference materials have significant two interfering elements, there are a larger number of critical
concentrations of the interfering element, the corrected curve is points, but the procedure is the same.
clearly established by the majority of the points. It becomes a
simple matter to determine the corrections required to bring the 10.6.3.6 The trial-and-error process should be repeated until
points off the curve onto it, and these corrections can be the fit of points to a smooth curve is comparable to the
divided by the concentrations of the interfering element to experimental error, which includes both the spectrochemical
determine correction factors. These will be approximately the precision and the reliability of the chemical values.
same, and a value should be adopted that is a weighted average, 10.6.4 Frequently, reference materials for calibration will
giving the strongest weighting to the standard reference mate- have been prepared by mixing powders or solutions, as in d-c
rial with the highest concentration of the interfering element. arc work or for the analysis of solutions or oils. Then it is a

13
 E 158
simple matter to synthesize several reference materials specifi- 10.9.1 For the intensity correction method, determine the
cally to confirm an interference and to calculate the correction concentration of the interfering element and multiply by the
factor. A reference material with low concentrations of the correction factor. Subtract the resulting correction from the
analytical element and the suspected interfering element is observed intensity ratio of the analytical element. Convert the
used to prepare additional reference materials with intermedi- corrected intensity ratio into concentration from the corrected
ate and high concentrations of the interfering element. If the analytical curve.
points for these additional reference materials fall above the 10.9.2 For the concentration correction method, convert the
point for the original standard, the interference is confirmed measured intensity ratio for the analytical element into “appar-
and the correction factor is calculated easily. ent concentration” from the corrected curve. Determine the
10.7 Computer Calculation of Interference Corrections—A concentration of the interfering element and multiply by the
computer can be programmed to calculate interference correc- correction factor. Subtract the result from the apparent concen-
tion factors, to obtain the best least squares fit to a smooth tration to obtain the corrected concentration.
curve. In its most sophisticated form it can do this directly from
10.10 It must be emphasized that this discussion of inter-
the measured intensity ratios, if the program includes a library
ference corrections assumes linear responses for both analyti-
file that permits the concentrations of all elements in all the
cal and interfering lines. If the analytical line self-absorbs at
calibration standards to be entered into memory. Calculations
higher concentrations, interference corrections using either
will be made much faster and more accurately than by manual
intensity ratios or concentrations will produce too great a
procedures. The use of a computer is particularly desirable
correction in the self-absorbed region. If the interfering line
when there are two or more interfering elements.
self-absorbs in the absorbing region, the interference correc-
NOTE 7—This can be dangerous because the computer may produce a tions will be too small, whichever method of correction is
beautiful fit of points to a curve from interferences that are not real. employed. Computers can be programmed to apply second-
Therefore it is essential to confirm the interference, as explained in 10.7, order corrections but the results are not likely to be accurate, so
and as illustrated for the 4057.82 Å lead line. Otherwise the analyses may
be inaccurate.
that every effort should be made to use only analytical lines
with linear responses and with linear corrections.
10.8 Interference Corrections Using Concentration
10.10.1 If the correction depends on the product of the
Ratios—Interference corrections are made in exactly the same
concentrations of both the analytical and interfering elements,
way when using concentration ratio calculations. Stated con-
an iterative solution can be programmed into a computer.
centration ratios replace stated concentrations. The same con-
centration correction factors should apply, but the corrections
11. Keywords
themselves will have two more zeros after the decimal point.
10.9 Determining an Element with Interference in an Un- 11.1 analytical curve; background correction; concentration
known Specimen—To calculate the corrected concentration for calculations; concentration ratio plots; concentrations; intensi-
an element with a known interference, once the corrected ties; intensity standard; interferences; internal standard
analytical curve has been prepared, proceed as follows for the method; optical emission spectrograph; optical emission spec-
two methods of correction: trochemical analysis; optical emission spectrometer

The American Society for Testing and Materials takes no position respecting the validity of any patent rights asserted in connection
with any item mentioned in this standard. Users of this standard are expressly advised that determination of the validity of any such
patent rights, and the risk of infringement of such rights, are entirely their own responsibility.

This standard is subject to revision at any time by the responsible technical committee and must be reviewed every five years and
if not revised, either reapproved or withdrawn. Your comments are invited either for revision of this standard or for additional standards
and should be addressed to ASTM Headquarters. Your comments will receive careful consideration at a meeting of the responsible
technical committee, which you may attend. If you feel that your comments have not received a fair hearing you should make your
views known to the ASTM Committee on Standards, at the address shown below.

This standard is copyrighted by ASTM, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.
Individual reprints (single or multiple copies) of this standard may be obtained by contacting ASTM at the above address or at
610-832-9585 (phone), 610-832-9555 (fax), or service@astm.org (e-mail); or through the ASTM website (www.astm.org).

14