This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles

for the
Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.

Designation: E168 − 16 (Reapproved 2023)

Standard Practices for

General Techniques of Infrared Quantitative Analysis1
This standard is issued under the fixed designation E168; 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 (´) indicates an editorial change since the last revision or reapproval.
This standard has been approved for use by agencies of the U.S. Department of Defense.

1. Scope E932 Practice for Describing and Measuring Performance of

1.1 These practices cover the techniques most often used in Dispersive Infrared Spectrometers
infrared quantitative analysis. Practices associated with the E1252 Practice for General Techniques for Obtaining Infra-
collection and analysis of data on a computer are included as red Spectra for Qualitative Analysis
well as practices that do not use a computer. E1421 Practice for Describing and Measuring Performance
of Fourier Transform Mid-Infrared (FT-MIR) Spectrom-
1.2 This practice does not purport to address all of the eters: Level Zero and Level One Tests
concerns associated with developing a new quantitative E1655 Practices for Infrared Multivariate Quantitative
method. It is the responsibility of the developer to ensure that Analysis
the results of the method fall in the desired range of precision
and bias. 3. Terminology
1.3 The values stated in SI units are to be regarded as 3.1 For definitions of terms and symbols, refer to Terminol-
standard. No other units of measurement are included in this ogy E131.
standard.
1.4 This standard does not purport to address all of the 4. Significance and Use
safety concerns, if any, associated with its use. It is the 4.1 These practices are intended for all infrared spectrosco-
responsibility of the user of this standard to establish appro- pists. For novices, these practices will serve as an overview of
priate safety, health, and environmental practices and deter- preparation, operation, and calculation techniques. For experi-
mine the applicability of regulatory limitations prior to use. enced persons, these practices will serve as a review when
Specific hazard statements appear in Section 6, Note A4.7, seldom-used techniques are needed.
Note A4.11, and Note A5.6.
1.5 This international standard was developed in accor- 5. Apparatus
dance with internationally recognized principles on standard- 5.1 The infrared techniques described here assume that the
ization established in the Decision on Principles for the equipment is of at least the usual commercial quality and meets
Development of International Standards, Guides and Recom- the standard specifications of the manufacturer. For dispersive
mendations issued by the World Trade Organization Technical instruments, also refer to Practice E932. For Fourier Transform
Barriers to Trade (TBT) Committee. and dispersive instruments, also refer to Practices E1421 and
E932 respectively, and for microanalysis with these instru-
2. Referenced Documents ments see Practice E334.
2.1 ASTM Standards:2 5.2 In developing a spectroscopic method, it is the respon-
E131 Terminology Relating to Molecular Spectroscopy sibility of the originator to describe the instrumentation and the
E334 Practice for General Techniques of Infrared Micro- performance required to duplicate the precision and bias of a
analysis method. It is necessary to specify this performance in terms
that can be used by others in applications of the method.
1
These practices are under the jurisdiction of ASTM Committee E13 on
Molecular Spectroscopy and Separation Science and are the direct responsibility of 6. Hazards
Subcommittee E13.03 on Infrared and Near Infrared Spectroscopy.
Current edition approved Jan. 1, 2023. Published January 2023. Originally 6.1 Users of these practices must be aware that there are
approved in 1964. Last previous edition approved in 2016 as E168 – 16. DOI: inherent dangers associated with the use of electrical
10.1520/E0168-16R23.
2
instrumentation, infrared cells, solvents, and other chemicals,
For referenced ASTM standards, visit the ASTM website, www.astm.org, or
and that these practices cannot and will not substitute for a
contact ASTM Customer Service at [email protected]. For Annual Book of ASTM
Standards volume information, refer to the standard’s Document Summary page on practical knowledge of the instrument, cells, and chemicals
the ASTM website. used in a particular analysis.

Copyright © ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. United States

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
1
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
7. Considerations for Quantitative Infrared account for the shift of the absorbance maximum. The question
Measurements arises whether it is preferable to measure absorbances at fixed
7.1 Quantitative infrared analysis is commonly done with wavenumber locations or at the observed maximum of the
grating, filter, prism, or interferometer instruments. The fol- analytical band. The best approach is empirical testing of both
lowing guidelines for setting up an analytical procedure are the fixed point and the tracking methods of evaluation.
appropriate: 7.1.6 Whenever possible, working directly in absorbance is
7.1.1 Always operate the instrument in the most stable and preferable. That is, either the instrument or associated data
reproducible conditions attainable. This includes instrument processor makes the necessary conversion from transmittance
warm-up time, sample temperature equilibration, and exact to absorbance. If spectra cannot be obtained in absorbance,
reproduction of instrument performance tests for both stan- then Eq A12.1 and A12.2 in Annex A12 can be used to convert
dards and samples. After calibration, use equivalent settings for the data.
analyses. For all infrared instruments, refer to the manufactur- 7.1.7 Use spectral regions offering the most information on
er’s recommendations for the instrument settings. After the analyte. Select analytical wavenumbers where the compo-
calibration, use these same settings for analysis. nent has a relatively large absorptivity. In addition, other
7.1.2 The absorbance values at analytical wavenumbers analytes should have minimal effect on the measured absor-
should fall within the acceptably accurate range of the particu- bance.
lar spectrometer used. In general, a single absorbance measure- 7.1.8 The performance of the spectrometer should be suffi-
ment will have the best signal-to-noise ratio when it is in the ciently good to give adequate linearity of response for the
range from 0.3 to 0.8 absorbance units (AU) (1).3 The desired range of concentrations. The signal-to-noise ratio, S/N,
sensitivity of Fourier transform (FT-IR) spectrometers is such should be acceptable for the desired precision.
that lower absorbance values can be used quite effectively, 7.1.9 Select analytical wavenumbers such that the linearity
provided that the baseline can be estimated accurately (see of the absorbance-concentration relationship is least affected
Section 12). Absorbances greater than 0.8 AU should be by molecular interaction, dispersion in refractive index, and
avoided wherever possible because of the possibility of spectrometer nonlinearity.
instrumentally-caused non-linearity, both for dispersive (2) and
FT-IR (3,4) spectrometers. Variation of the concentration and 8. Theory for a Single-Compound Analysis
sample path length can be used to adjust absorbance values into 8.1 Quantitative spectrometry is based on the Beer-
the optimum range. When multiple components are determined Bouguer-Lambert (henceforth referred to as Beer’s) law, which
in a particular sample, it is acceptable to use absorbance values is expressed for the one component case as:
outside the optimum range, (5) however, absorbances greater A 5 abc (1)
than 1.5 AU should be avoided (2-4). Weaker absorption bands
of high concentration components may be selected to provide where:
absorbance values within the optimal range. A = absorbance of the sample at a specified wavenumber,
7.1.3 The most accurate analytical methods are imple- a = absorptivity of the component at this wavenumber,
mented with samples in solution. With liquid samples that are b = sample path length, and
not exceptionally viscous, best results are obtained if the cell is c = concentration of the component.
not moved after the first sample is introduced into the instru- Since spectrometers measure transmittance, T, of the radia-
ment (the fixed-cell method). The reason is that sample cell tion through a sample, it is necessary to convert T to A as
position is difficult to reproduce accurately by insertion into follows:
typical cell holders. Suitable fittings and tubes can be attached
P
to the cell to allow sample changing in a flow-through manner. A 5 2logT 5 2log (2)
P0
When it is not practical to use a flow-through cell, the cell
should fit tightly in the holder so that lateral and tilting motions where:
are restricted. P0 = input radiant power at the sample, and
7.1.4 Unless there is reason to suspect deposition on or P = radiant power transmitted through the sample.
contamination of the cell from the samples, it is generally
preferable to wash out the current sample with the next sample, 9. Calibration for a Single-Component Determination
if sufficient sample is available. The volume of sample used to
9.1 Proper sample preparation is essential to quantitative
flush the cell should be at least five times (and preferably more,
analysis. See Annex A4.
for example, 20 times) the volume between the sample inlet
9.1.1 Quantitative analysis has two distinct parts: calibra-
and cell exit points.
tion and analysis. For a simple one-component analysis, select
7.1.5 For some bands, the wavenumber of the maximum
an appropriate solvent that is essentially free from interfering
absorbance changes as a function of concentration. Similarly,
absorptions at the analytical wavenumber.
the position of the baseline points may change with concen-
9.1.2 For calibration, measure the absorbances, A, of the
tration. Selection of baseline points must be done carefully to
analyte solutions at several known concentrations, c.
Absorptivities, a, are then calculated, using Eq 1 with the
3
The boldface numbers in parentheses refer to the list of references at the end of baseline corrections as described in Sections 12 – 14.
these practices. Alternatively, the absorbances, A, of a single solution in several

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
2
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
cells of different, but accurately known, path lengths may be independent equations containing n absorbance measurements
measured; however, interaction effects will not be elucidated in at n wavenumbers are necessary. This is expressed for constant
this fashion. path length as follows:
9.1.3 Calculate the average of the several a values for future A 1 5 a 11bc1 1a 12bc2 1····1a 1n bcn (4)
use, or draw an analytical working curve by graphing absor-
bance versus concentration for a constant path length as A 2 5 a 21bc2 1a 22bc2 1····1a 2n bcn
demonstrated in Fig. 1. Use the linear part of the curve to
calculate a. The calculation of a where curvature is present will ·· ········ ······
be discussed in 18.1 and 18.2.
·· ········ ······ ······
NOTE 1—In practice, the calibration curve may not have a y intercept of
zero. This could be due to a variety of factors including, but not limited to, ·· ········ ······ ······
incompletely resolved analyte bands, reflection losses, and solvent inter-
ferences. It is important that the method used to calculate the calibration A i 5 a i1 bc1 1a i2 bc2 1····1a inbcn
curve not force the y intercept to be zero.
where:
9.1.4 For analysis, dissolve the unknown in the solvent,
measure the absorbance, A, and determine the concentration, c, Ai = total absorbance at wavenumber i,
ain = absorptivity at the wavenumber i of component n,
of the analyte graphically or by calculation. Convert this
b = path length of the cell in which the mixture is sampled,
concentration in solution to the concentration in the unknown
and
sample. cn = concentration of component n in the mixture.
9.1.5 Both analysis time and chance of error are less if the
concentrations of the unknowns and the cell path length are 10.2 During calibration, concentrations cn are known, and
kept the same over a series of analyses, and the concentrations baseline corrected absorbances A are measured. The experi-
of the calibration solutions have bracketed the expected high mental absorptivity-path length products ainb are then calcu-
and low values of the unknown solutions (6, 7). lated (see Note 2). During analysis, the absorptivity-path length
products ainb are known, and the absorbances A are measured.
10. Theory for Multicomponent Analysis The unknown concentrations are then calculated (see Section
10.1 Beer’s law is expressed for a mixture of n indepen- 17). Therefore, accurate calibration generally requires that
dently absorbing components at a single path length and single experimental absorptivity values be obtained from at least n
wavenumber as: standards. The following requirements must be met:
10.2.1 The number of standards must be equal to or greater
A 5 a 1 bc1 1a 2 bc2 1······1a n bcn (3) than the number of analytes, n, and
Eq 3 defines an absorbance at a wavenumber as being due to 10.2.2 The number of analytical wavenumbers, i, must be
the sum of the independent contributions of each component. equal to or greater than the number of independent
In order to solve for the n component concentrations, n components, n.
NOTE 2—All absorbance conversions use transmittance (that is, the
decimal value), not percent transmittance. Regardless of form (that is,
decimal or percent), the term transmittance refers to the term P/P0 of Eq
2, and should not be called transmission. (See Terminology E131).
10.3 The first requirement allows the analyst to use more
than the minimum number of standards. Over-determination of
standards permits error estimation in the analytical result. The
second requirement allows the use of more than the minimum
number of peaks for specifying a chemical system, where at
least one distinctive band is selected for each component
(7-10).
10.4 The procedures used in multicomponent analysis will
be discussed further in the following section which is also an
introduction to general solution phase analyses.

11. Multicomponent Solution Analysis

11.1 For the quantitative analysis of mixtures, Eq 4 is
applicable. The absorptivities ain of the n components of the
mixture at the ith analytical wavenumber are determined from
absorbance measurements made on each component taken
individually. These absorbances must be measured under
conditions (sample path length, temperature, pressure, and
solvent) identical to those used for the unknowns, and they
FIG. 1 An Analytical Working Curve should be corrected for baselines as discussed in Sections 12 –

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
3
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
14. Absorbance measurements are made with concentrations of
the analyte bracketing the amounts expected in the unknown
samples.
11.2 Where possible, prepare samples as dilute solutions
and place in cells of appropriate path lengths (typically 0.2 to
1.0 mm). Use lower concentrations in longer path length cells
rather than higher concentrations in shorter path length cells to
obtain absorbance values in the 0.3 to 0.8 range. Lower
concentrations will minimize nonlinear effects due to disper-
sion (that is, change of refractive index with wavenumber).
Where freedom from intermolecular effects is uncertain or
where intermolecular effects are known to be present, calibra-
tion must be based on measurements taken from synthetic
mixtures of all components as described in 15.1.2.
11.3 Dissolve a known weight of a pure component in a
suitable infrared solvent. Measure the absorbance at all ana-
lytical wavenumbers and correct for baselines as discussed in
Sections 12 – 14. Repeat this procedure for several concentra-
tions covering the range of concentrations expected in the
samples to be analyzed, remembering that concentrations of
components must be linearly independent. Plot absorbance
versus concentration. Similarly, construct analytical curves for
this component at each of the other analytical wavenumbers. FIG. 2 A Zero-Absorbance Baseline
Repeat this procedure for each of the n components. Thus,
there are i plots for each component, or a total of i × n 14. Baseline Method (7)
analytical curves, each yielding one of the values of ainb.
14.1 The cell-in-cell-out technique was the method of
11.4 The number of standard mixtures required is at least choice for early single-beam infrared instruments. After the
equal to n, the number of components. For each analytical advent of double-beam dispersive spectrometers, the baseline
wavenumber, there will be a set of at least n equations in n method has been the method of choice. Portions of the data
unknowns. The n sets of equations can be solved directly for around the base of the bands are picked as baseline references.
the values of ainb. If more than n synthetic mixtures are used There are two common variations.
as standards, a least-squares procedure can be used to calculate 14.2 When one baseline point is chosen, the value of an
the values of ainb. To repeat, in order to obtain information absorbance minimum, A2, is subtracted from the absorbance
about errors, at least one more mixture than the number of maximum, A1, as demonstrated in Fig. 3. The point of
analytes is needed. minimum absorbance is adjacent to or at least in the vicinity of
the band under evaluation.
12. Baselines in General
14.3 Two points may be needed if the band of interest is
12.1 Any quantitative method depends on the choice of a superimposed on a sloping background. Manually a line is
reproducible baseline. The correction of raw data for baseline drawn from one side to the other as in Fig. 4. The absorbance
absorbance is important in some methods. The guiding factor of the band is calculated as the value at the peak maximum A1
in baseline selection is the reproducibility of the results. minus the baseline absorbance minimum A23. An inappropriate
Methods used for drawing baselines with computerized instru- choice of baseline in this situation may have deleterious effects
ments are similar in most ways to those for data recorded on on the accuracy of the final calculation.
chart paper. Where differences exist, they will be explained in
Annex A1. NOTE 3—The above baseline correction procedure should be performed
only if the spectrum is plotted in absorbance units. When the spectrum is
plotted in transmittance, the two baseline transmittances and the transmit-
13. Single Wavenumber Measurement tance at the analytical wavenumber should be converted to absorbance.
The corrected baseline absorbance can be calculated by Eq A12.1 in
13.1 A technique known as the “cell-in-cell-out” method is Annex A12. Conversion to absorbance is required because a sloping linear
often used in single-beam infrared work. In this method, a baseline in transmittance becomes curved in absorbance.
blank (that is, solvent in cell, potassium bromide (KBr) pellet,
or other substrate) is measured at a fixed wavenumber and then 15. Nonsolution Analyses
the analyte readings are recorded (7). In the simplest cell-in- 15.1 Liquids:
cell-out method, a zero absorbance baseline is used (see Fig. 15.1.1 Analyzing a liquid mixture without the use of a
2). If the spectrum cannot be obtained in absorbance, the diluting solvent is sometimes complicated by intermolecular
absorbance is calculated as in Eq A12.1 where T2 = 1.0 and T1 forces. An absorption band may undergo intensity changes or
= transmittance at the analyte wavenumber (1, 6) (see Note 2). frequency shifts, or both, relative to the same absorption band

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
4
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
chosen for the analysis and substitute them (along with the
known concentrations) in Eq 4. Solve for the absorptivity-path
length products, ainb directly from the set of n simultaneous
equations, or use a multivariant method (see Annex A8) if
sufficient data are available.
15.1.3 If the concentrations in the unknowns vary widely,
calculation of a second set of the ainb products is recom-
mended. A second set may be necessary due to the presence of
intermolecular influences, and the differences in the values of
the absorptivities thus determined will indicate the extent of
these influences.
15.1.4 A single set of absorptivities may not suffice to
analyze mixtures throughout all possible concentration ranges
of the components, in which case, narrowing the range of
concentrations is recommended.
15.1.5 Since the ainb products are calculated directly in this
procedure, it is not necessary to plot analytical curves.
15.2 Solids:
15.2.1 For cast films, pressed films, or pellets, follow the
same general procedure as for liquids (see 15.1). Measure the
thickness of each film and apply a proportional correction for
deviations from standard thickness.
FIG. 3 A One-Point Baseline
NOTE 4—The spectra of films and pellets can be complicated by the
presence of a fringe pattern. For pellets and films, follow the suggestions
in A4.5.1.2 and Note A5.1, respectively. A fringe pattern is undesirable
because analyte absorbance values can be altered by its presence.
15.2.2 In cases where all components of a mixture are
determined to a total of 100 %, it is usually sufficient to
determine only the ratios of absorbances. In such cases, it is not
necessary to know the thickness of the sample layer; it is only
necessary to know the ratio of the components. However, a
knowledge of the thickness is needed to determine the presence
of impurities because the total then will be less than 100 %.
15.2.3 The above procedure for films is also used with
powders prepared as mulls. Measurement of thickness can be
accomplished by an internal standard technique as described in
A4.4.2. This involves the addition to the sample of a known
weight ratio of a compound having an absorption band of
known absorptivity that does not overlap the bands of the
sample.
15.2.4 When powders are measured as pressed plates or
pellets, analytical curves are prepared in the same manner as
solutions, see Sections 9 and 11.
15.3 Gases:
15.3.1 All calibration measurements for a given analysis
must be made at a fixed total pressure. This pressure must be
equal to the total pressure employed in the analysis. An
FIG. 4 A Two-Point Baseline analysis may be set up in either of two ways:
15.3.1.1 Method 1—A fixed sample pressure is established
of the component in solution. The absorbance contribution of a that is a fraction of the total pressure obtained by addition of a
component in a mixture can seldom be calculated from its nonabsorbing diluent gas.
absorbance measured in the pure state. It is desirable to 15.3.1.2 Method 2—A fixed sample pressure is used as the
determine the absorptivities from known mixtures having total pressure. Analytical curves are prepared by introducing a
proportions near those of the samples. pure component at various measured pressures which bracket
15.1.2 Prepare mixtures having known concentrations of the the expected component pressures in the sample. A diluent gas
various components covering the expected ranges. Measure is then added to bring the total pressure up to the established
baseline corrected absorbances at each of the wavelengths value.

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
5
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
15.3.2 In Method 2, the analytical curve preparation does
not allow for the possibility of band broadening for different
components. This factor is more properly addressed by follow-
ing Method 1 where the same diluent gas is employed for
sample preparation and calibration. Low molecular weight
gases frequently produce very strong, sharp absorption fea-
tures. Addition of a diluent gas and use of pressure less than
atmospheric may be necessary. Absorbances are measured for
each standard at the wavenumbers selected for analysis. Where
possible, integrated absorbances (see Annex A3) are preferred
to offset the effect of small pressure variations. The absor-
bances are plotted against the partial pressures (or mole
fractions) to produce analytical curves.
16. Difference Method
16.1 Spectral subtraction using a computer is a common
practice in qualitative infrared analysis. This technique is also
used to perform quantitative infrared analyses. The advantage
of spectral subtraction (the difference method) is that small
concentration differences can be measured with greater accu-
racy than is possible on superimposed bands.
16.2 A generalized procedure follows and is illustrated in
Fig. 5. All spectra are obtained using samples of well charac-
terized path length and concentration. Fig. 5(c) shows the
spectrum of Z, an unknown mixture containing components X
and Y. Using a subtraction routine, the spectrum of X is
removed using the isolated, in this case higher, wave-number
bands of X as a guide (11). The concentration of Y is
ascertained from Fig. 5(d) by reference to an analytical curve
or by calculation as described in 9.1.3.
16.3 The same result is achieved with a noncomputerized
double-beam spectrometer by placing sample X in the refer-
ence beam, and the unknown mixture in the sample beam. If
the sample and reference are in solution, a variable path length
cell can be used in the reference beam to remove spectral
contributions due to X (7, 12).
17. Calculation Methods
17.1 Matrix Inversion:
17.1.1 After the values of the ainb products have been
FIG. 5 An Example of Difference Spectroscopy
determined for a given set of n components, according to 10.2,
substitute the numerical values into Eq 4. Solve the n equations
for concentrations, cn, in terms of the baseline corrected 17.2 Matrix inversion is a convenient method to calculate
absorbances, An, by matrix inversion (6). The inverted equa- concentrations from the simultaneous equations presented in
tions will have the following form: Eq 4. Programs for solving simultaneous linear equations using
C 1 5 A 1 F 11 1A 2 F 12 1····1 A n F 1n (5) matrix-inversion techniques are available on many program-
mable calculators and computers and are contained in most
C 2 5 A 1 F 21 1A 2 F 22 1····1A n F 2n commercial quantitative analysis programs. Classical least
squares regression (CLS) is simply a sophisticated method of
·· ····· ····· ···· ·····
matrix inversion (see Annex A8).
·· ····· ····· ···· ·····
18. Correction for Curvature in Beer’s Law Plots
·· ····· ····· ···· ····· 18.1 In some cases, the analytical curve of one or more
analytes of a mixture will exhibit curvature to such an extent
C n 5 A 1 F n1 1A 2 F n2 1····1A n F nn
that the value of the slope may differ significantly between low
where Fin are the inverted coefficients. Thereafter calculation and high concentrations. Two methods are acceptable: a
of individual sample concentration is simply done by substi- non-linear regression using a computer or graphical method as
tuting the measured absorbance values, An, in the equations. immediately explained. If the graphical method (see 9.1.3) is

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
6
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
used, and if the concentrations of analytes fall in the linear and tions from Beer’s law. Quantitative bias depends upon mini-
low range, then the values of the slope for the linear range can mizing determinate error.
be used. However, if the concentration is in the higher range, a 19.3 Indeterminate, or random, error arises from uncontrol-
correction is necessary. The following method is recom- lable variables, and limits the precision with which measure-
mended: ment can be made. Often the major indeterminate errors are
18.1.1 The concentration of the component under consider- introduced by variation in sample positioning and errors in
ation ranges in the sample between c1 and c2 in Fig. 1. Draw a determining the baseline. However, if these are held constant,
straight line between A1 and A2. The slope of this line is the the major contributing indeterminant error frequently is detec-
value of ainb that is used in Eq 4. The intercept of this line with tor noise, which is usually independent of signal. Therefore,
the absorbance axis yield the value of a correction term, A0, the noise in transmittance units is independent of the amount of
which must be subtracted from the measured absorbance of the light reaching the detector. For a review of the sources of noise
sample at the analytical wavenumber of the analyte. This in Fourier transform instruments, see Ref (11) and Practice
subtracted result is substituted for A2 in Eq 4 at this analytical E1421.
wavenumber. If the concentration of the component under
scrutiny should happen to fall outside the range c1 to c2, it will 19.4 For quantitative infrared spectrometry, the operative
be necessary to repeat the above procedure to determine the equation for determining concentration from transmittance
slope and intercept for the new concentration range. measurements is Beer’s law as follows:
18.2 In some binary mixtures, pure bands representing the A 5 2logT 5 abc (6)
individual components are not present. However, single bands c 5 A/ab (7)
or groups of bands, as intensities or area, can be ratioed and
To determine the effect of random error (in the measurement
plotted to the known concentrations (13). These calibration
of transmittance) (1, 12, 15) on the concentration, it is
curves are almost always curved, but as explained in Ref. (13),
necessary to calculate the partial derivative as follows:
curved absorbance/concentration plots are not a problem since
numerous computer programs are available for non-linear δc 2loge 20.434
5 5 (8)
regression analysis. δT abT abT

19. General Considerations for Statistical Evaluation The standard deviation of the concentration sc can be given
by:
19.1 The statistical evaluation of experimental data and the
parameters necessary for reporting statistical confidence are
described in this section and in Annex A6. The reliability of an
sc 5 S D
0.434
abT
sT (9)

experimentally measured quantity is an important factor which where sT is the standard deviation of the transmittance
must be considered in evaluating any experimental technique. measurement. The relative standard deviation of the concen-
This reliability can be described by two terms: precision and tration is:
bias. The precision of a technique refers to the reproducibility
of replicate measurements; the bias represents the degree to
which the measured quantity approaches the true value. The
sc
c
5 S
0.434
logT DS DsT
T
(10)

sources of experimental error limiting bias or precision, or and the standard deviation of the transmittance is calculated
both, are broadly classified as determinate or indeterminate from Eq A6.6 for a series of n measurements of T. sT can be
error (1, 14, 15). determined from the noise in the 100 % line since generally sT
19.2 Determinate error is systematic error which can be will be independent of T.
attributed to definite causes. In quantitative infrared analyses,
determinate error may arise from problems such as optical 20. Keywords
misalignment, photometric inaccuracy, stray radiant power, 20.1 infrared spectroscopy; molecular spectroscopy; quan-
poor spectral resolution, improper sample handling, or devia- titative analysis

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
7
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
ANNEXES

(Mandatory Information)

A1. BASELINE PROCEDURES FOR COMPUTERIZED INSTRUMENTS

A1.1 Obtaining a Good Spectrum for Baseline Procedures the sample spectrum. This paragraph describes in general how
subtraction works with both FT-IR and dispersive spectra.
A1.1.1 There are two ways to get good (FT-IR) spectra. For
the first method, three steps are necessary. 1) Obtain both A1.1.3 For dispersive, optical-null instruments, the selec-
single-beam background and single-beam sample spectra. 2) tion of instrumental settings or mode (for example, resolution,
Ratio the single-beam sample to the single-beam background scanning region, etc.) are based on sample characteristics and
spectrum. This provides a spectrum in transmittance and the absorbance of the functional group being measured.
requires conversion to absorbance. 3) Convert to absorbance A1.1.4 In general, for both spectrometer types, spectral data
by computing the negative logarithm of the transmittance are collected by the cell-in-cell-out method of 14.1. A baseline
spectrum. The background can be that of the open beam, or an method is then used to obtain the actual quantitative data.
aperturing device, or an accessory, such as an ATR or gas cell These methods are demonstrated in Figs. 2-4.
or a liquid cell containing solvent used to dissolve the analyte.
A1.2 Calculation Procedure
NOTE A1.1—Dispersive duel-beam instruments perform the above by A1.2.1 The calculation of data with one baseline point is
using the reference beam to obtain a background simultaneously. Hence, discussed in 14.2.
the reference beam should contain similar beam limiting accessories as
used in the sample beam. In some cases, for example with a gas cell, the A1.2.2 Automatic computation of peak absorbance with a
above approach is impractical. The alternative is to run spectra of the two-point baseline is more subject to error. The calculations are
sample and the empty accessory separately, then using the computer based on the point-slope method, where the hypotenuse of a
software to subtract the empty accessory from the sample. Keep in mind right triangle is the desired sloping baseline as shown in Fig. 4.
that these spectra must be in absorbance.
The slope of the baseline may be either positive or negative.
A1.1.2 The second method similar to A1.1 is frequently The peak absorbance is the result of the following:
used when the spectrum of the sample material contains ~ A 3 2 A 2! ~ w 1 2 w 2!
extraneous absorption features (for example, solvents or impu- A 5 ~ A 1 2 A 2! 2 (A1.1)
~ w 3 2 w 2!
rities). In this approach, the single-beam spectrum of the
sample and the single-beam spectrum of the solvent or impu- where:
rity are each ratioed against the single-beam background A = corrected absorbance of the peak at w1,
spectrum. Both transmittance spectra are converted to absor- A1 = uncorrected absorbance at w1,
bance. The absorbance spectrum of the solvent or suspected A2 = baseline absorbance point at the lower wavenumber w2,
impurity is then scaled by multiplying it by a factor chosen to and
minimize all spectral features caused by the solvent or impu- A3 = baseline absorbance point at the higher wavenumber
rity. The impurity or solvent spectrum is then subtracted from w3.

A2. GENERAL CONSIDERATIONS FOR BAND AREA

A2.1 All data should be expressed in absorbance as a band area is found to be more accurate than peak-height
function of wavenumber. measurements because one is, in effect, averaging multipoint
data.
A2.2 Band shape changes can cause peak-height data to be
nonlinear. Band area, however, may remain essentially unaf- A2.4 When integrated area is used for quantitative analyses,
fected by the changes in shape of the band because band area the reliability of the results frequently depend on the baseline
is a function of the total number of absorbing centers in the treatment selected. The accuracy by band area is often im-
sample. If the shape change is caused by changes in intermo- proved by limiting the range of absorbances. The wings
lecular forces, even band area may not be linear. contribute very little signal while contributing substantial
A2.3 Band area is calculated by integrating across band- uncertainties to the total area. A useful guideline is to limit the
width. Band area is advantageous when band shape undergoes integration limits to absorbance values which are no smaller
change as a function of increasing concentration. Frequently, than 20 % to 30 % of the peak absorbance.

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
8
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)

A3. CALCULATION OF BAND AREA

A3.1 In reference to Fig. A3.1, when no baseline points are where I0 is given by Eq A3.1.
used for an area calculation, the area between lower and upper
wave-number limits is the following: A3.3 If a two-point baseline treatment is used with absor-
bances AW2 at wavenumber w2 and Aw3 at wavenumber w3, as
I 5 A w4 ∆1A w411 ∆1···1A w521 ∆ (A3.1) shown in Fig. A3.3, the formulation is as follows:
where: I 2 5 I 0 2 I 2b (A3.3)
I = integrated absorbance (area),
∆ = sampling interval in wavenumbers,
A = absorbance measured at the designated interval,
I 2b 5 F (S
np

j51
A w2 1
~ A w3 2 A w2 ! ~ w 4 2 w 2 1j∆ !
~ w 3 2 w 2! DG ∆ (A3.4)

w4 = lower wavenumber limit, and

w5 = upper wavenumber limit.
The number of points in the sum is np = (w4 − w5)/∆. and I0 is given by Eq A3.1.
NOTE A3.1—The algorithms above are not the most accurate, but as ∆
A3.2 In reference to Fig. A3.2, for a one-point baseline becomes smaller, all methods (that is, trapezoidal and Simpson’s rule)
treatment w2 with a corresponding absorbance Aw2, the area I1 approximate the same value.
is as follows:
I 1 5 I 0 2 I 1b (A3.2)

I 1 5 I 0 2 ~ A w2 ! ~ w 5 2 w 4 !

I 1 5 I 0 2 @ ~ ∆ w2 ~ np 2 1 !! #

FIG. A3.1 Band Area With Zero-Absorbance Baseline

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
9
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)

FIG. A3.2 Band Area With a One-Point Baseline

FIG. A3.3 Band Area With a Two-Point Baseline

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
10
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)

A4. SAMPLE PREPARATION

A4.1 Where possible, solution techniques are recom- A4.4.3 Break up the mixture and distribute it over the
mended. However, other methods are discussed. mortar surface by gentle grinding with the pestle. Rub to an
extremely fine powder by vigorous back and forth grinding
A4.2 Liquids—Liquids may be measured either neat or until the mixture becomes caked and takes on a smooth, glossy,
preferably, as solutions in a sealed cell of suitable path length light-reflecting surface.
(15).
A4.4.4 Add one drop of mulling agent (see Note A4.5),
A4.3 Solution Techniques: grind with a rotary motion, using a smooth, hard mortar and
pestle, until all of the mixture is picked up and suspended. The
A4.3.1 For liquids and solids, the following procedures are
suspension should be viscous, not fluid.
recommended but may not be all encompassing:
A4.3.2 Choose a solvent that has minimal absorbance at the NOTE A4.5—Mineral oil, fluorinated hydrocarbon, or appropriate sub-
stances can be used as the mulling agent. Matching the refractive indices
analytical wavenumbers and that will completely dissolve the as closely as possible gives best results.
sample.
A4.4.5 Transfer the sample to a clean, flat salt plate with a
A4.3.3 Measure a specified amount of sample into a volu- clean, rubber policeman or clean, plastic spatula.
metric flask or other suitable container, and add some of the
solvent to be used. The analyte must be completely dissolved. NOTE A4.6—A discussion of salt plates and cell-window materials is
covered in Practice E1252.
A4.3.4 Allow for temperature equilibration to the same A4.4.6 Cover with a second clean, flat plate; squeeze and
temperature as that used for the standards. rotate to obtain the desired thickness and to remove all trapped
A4.3.5 Make up to volume or weight with solvent and mix air.
thoroughly. A4.4.7 Visually observe the scattering of light passing
A4.3.6 Aqueous solutions can be analyzed using flat surface through the sample. As a guideline, the sample may appear
or circular ATR techniques. slightly hazy, but objects on the far side should be distinguish-
able.
NOTE A4.1—For other difficult solutes, such as elastomers and tars, it is
frequently more convenient to roughly weigh the solute in a suitable
container, add solvent from a graduated cylinder, dissolve, and run the
A4.5 Solids By Pressed-Pellet Technique (8, 12):
analysis. The concentration is then obtained by doing a percent solids A4.5.1 Particle size of the sample should be reduced to less
content on an aliquot from the remaining solution. than the analytical wavelength. This is relatively easy for many
NOTE A4.2—Solvent influence or overlap on the absorption bands to be
measured must be recognized and taken into account in the calculation.
samples without incurring any change in the sample. However,
NOTE A4.3—For best results, measure unknown sample solutions at polymorphic changes, degradation, and other changes may
concentrations that will place the analyte absorbances in the range of those occur during grinding. If such changes do happen and are not
used for calibration. controlled, the method will not be valid.
A4.5.1.1 Grinding conditions must be established for the
A4.4 Solids by Mull Techniques (8, 12):
particular sample type since grinding severity can affect
A4.4.1 The following is the recommended procedure for absorption-band intensity.
preparing solids by mull techniques; however, other methods A4.5.1.2 Weigh the preground sample and powdered potas-
also may be appropriate. Average particle size of the sample sium bromide (KBr) or other alkali halide in specified amounts
should be reduced to less than the analytical wavelength. The (sample weight should be about 1 % of the KBr weight; about
particle size reduction is relatively easy for many samples 350 mg KBr is appropriate for a 13 mm disk to avoid
without incurring any change in the sample. However, poly- interference fringes), and mix by hand in a mortar. This should
morphic changes, degradation, and other changes may occur be a mixing step rather than a grinding step.
during grinding. If such changes do occur and are not A4.5.1.3 Place the mixture in an appropriate evacuable die,
controlled, the method will not be valid. If changes are evacuate to at least 15 mm Hg and press at sufficient pressure
suspected, other techniques such as attenuated total reflectance to produce a transparent disk. Follow the manufacturer’s
(12) or diffuse reflectance (11) should be investigated. recommendation as to pressure for a particular die. Other
A4.4.2 Weigh slightly more than the minimum amount of methods, such as minipress cells are also used, but these
sample required, and then weigh the desired amount of an methods may not be as satisfactory due to crazing caused by
appropriate internal standard. Mix thoroughly. trapped air and water vapor.
NOTE A4.4—An appropriate internal standard (7, 11, 12) is a substance NOTE A4.7—Precaution: During pressing operations, place the die
that (a) exhibits a band in a suitable region of the spectrum, and as close symmetrically in the press. Otherwise, the die may forcefully slip out of
to the analyte’s wavenumber as possible, (b) is not present in the sample, the press, causing personnel injury or damage to surrounding equipment.
and (c) does not react chemically with or dissolve in the sample or mulling Some laboratories require safety shields in front of presses.
agent. An alternative to a separate internal standard is to use a band in the NOTE A4.8—Since the purity of alkali halide powders are not all the
sample that does not change as the moiety of interest is varied. This same, the same alkali halide powder should be used for the sample and
approach is very useful in polymer analyses. blank.

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
11
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
NOTE A4.9—If a steel-ball mill is used, clean, dry, rust-free stainless closed-loop pumping system of known volume. A known (or
steel vials and balls are recommended. A blank pellet (7) is made in the unknown) sample can be injected from a gas syringe into the
same manner as for the sample using the same ball-mill equipment, and
procedures. closed loop containing a diluent gas. The closed loop greatly
NOTE A4.10—Samples may undergo polymorphic changes or reactions reduces the time for mixing and helps to provide a homoge-
such as oxidation, ion exchange, or salt formation with the alkali halide neous sample for analysis.
material. These events may also invalidate a method.
NOTE A4.11—Precaution: Safety considerations require the use of
A4.6 Gaseous Samples (12, 15, 16, 17): proper pressure reduction valves for metering gases into a sample bomb.
A4.6.1 Take care that samples do not affect cell windows, Take precaution when mixing potentially hazardous or reactive gases.
mirror, or interiors. NOTE A4.12—All measurements for calibration and analysis under a
A4.6.1.1 Either fixed or variable path length cells can be given analytical system should be at a single, reproducible total pressure,
since the band intensity is a function of the partial pressure of the analyte,
used for gas analyses. Cells with path lengths from 1 cm to 20 the total pressure of the sample, and the total pressure within the cell and
m are readily available and other lengths can also be obtained. the temperature.
A4.6.1.2 Cells can be filled through the use of a vacuum NOTE A4.13—Mixing does not happen automatically. Attention must be
system coupled to a pressure gage for measurement of partial paid to assure complete transfer and equilibrium mixing.
pressures. NOTE A4.14—Calibration mixtures should be cross-checked by an
A4.6.1.3 An alternative method for filling and using long independent method.
path length cells is to have the cell incorporated into a

A5. PREPARATION OF POLYMERS

A5.1 Polymeric materials are studied by a variety of tech- rapidly to the desired value and released after a predetermined
niques; the choice depends on the physical properties of the time. The template and the pressed film are quickly quenched
particular polymer. Thermoplastics are hot pressed to form (and dried gently, if necessary). Film thickness is determined
films of a desired thickness. Cross-linked elastomers are often by averaging multiple micrometer measurements of the sample
prepared by microtoming, while others, particularly reinforced area in the infrared beam. These measurements can then be
materials, can be cold ground and pressed into a pellet using a used to calibrate the thickness with standard bands.
suitable sintering agent such as KBr or made into a mull using NOTE A5.1—Occasionally, a sinusoidal pattern is superimposed on a
a suitable suspending liquid such as mineral oil or fluorocar- spectrum. This is referred to as fringing. The fringe pattern is caused by
bons. Another method of forming a thin film is to cast from multiple internal reflections within the sample. This phenomenon can be
solvent. This is especially suitable for resins which either are reduced by placing fine (Grade 6/0) corundum paper between the first
sheet of aluminum foil and the backing plate. The corundum paper causes
not thermoplastic or break down at pressing temperatures. All a matte finish which scatters the reflected light diminishing the multiple
of these techniques can be successfully used for quantitative or internal reflections. This technique must be used cautiously because
semiquantitative analysis. If the above methods are not precision can be lessened by the matte finish (7).
satisfactory, either ATR (12) or diffuse reflectance (11) may be NOTE A5.2—Aluminum foil or other metal in contact with the molten
useful. polymer may need a mold-release agent. Many releasing agents contain
silicone polymers which can interfere with the peaks of interest. Wiping
A5.2 Preparation of polymers by the hot-pressing technique the surface with a solvent is sometimes effective, but use of cover plates
requires a suitable hydraulic press with heated platens capable coated with TFE-fluorocarbon or with a nonmigratory releasing agent is
of reaching the softening point of the polymer. In some cases, better.
water-cooled platens are also necessary to obtain the desired NOTE A5.3—For very thin films, templates are less useful, and
release-coated, flat metal plates can be used. Use smaller amounts of
results. The pressing zone for a number of common polymers polymer to achieve thinner films.
is given (18, 19, 20, 21). For quantitative results, the use of a NOTE A5.4—For quantitative determination of blended components, it
template with an opening to accommodate the infrared beam is is advisable to increase homogeneity by pressing a large sample, about 5
recommended. Typically templates from 0.3 mil to 6 mil g, into a film. Cut this film and stack the pieces so outer edges will lie on
(7.5 µm to 150 µm) are easily cut from metal shimstock. center portions, and then repress. Select smaller samples from this blend
Sufficient powdered or pellets of polymer to fill the opening is for final pressing.
placed in the template between two layers of aluminum foil NOTE A5.5—Pressing techniques must be used with caution when
investigating crystalline polymers, or polymers with other order-
which are then placed between two larger smooth backing producing structures. These polymers are temperature sensitive. Some
plates. A small amount of pressure is applied while the polymer problems can be avoided by annealing under controlled conditions before
heats to the desired temperature, then pressure is increased quantitative measurements are made.

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
12
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
A5.3 Microtoming —This technique is used for cross-linked A5.6 Polymers can be prepared as cast films. Selection of
and hard elastomers. Hand-driven and automated microtomes solvent is important since both solubility of the polymer and
are available where sample size and thickness are preselected volatility of the solvent affect the resultant film (23). A film can
for the analysis method. Temperature of the sample and knife be cast on the surface of an IR transparent material, or a
should be controlled within the plastic range of the material free-standing film can be cast on the surface of glass, metal,
(22). water, TFE-fluorocarbon, polyethylene, or mercury
(Warning—See Note A5.6.) The dried film is then mounted
A5.4 Friable polymers may be prepared for infrared exami-
nation by the pressed-pellet technique as in A4.5. Grinding at directly in the infrared beam. Uniform thickness is very
reduced temperatures allows the technique to be extended to difficult to achieve and best quantitative results are obtained by
soft or rubbery polymers. Avoid water contamination by using a calibrated, preselected internal-thickness band.
condensation. Quantitative results are obtained by one of NOTE A5.6—Warning: Mercury vapor is toxic. Exercise precaution
several methods: (a) using an internal standard band from one when mercury is used.
of the components, (b) adding a known quantity of an inert NOTE A5.7—Polymer crystallinity and phase separation of blends
commonly result as film drying progresses. Multiple areas should be
internal standard such as potassium thiocyanate, or (c) using a
sampled to test composition, and to provide average values.
known concentration of polymer in the pellet matrix and
measuring the thickness of the pellet.
A5.5 Friable polymers may also be prepared for infrared
examination by the mull technique as in A4.4. For quantitative
measurements, the best results are usually obtained by using an
internal reference band, but adding an internal standard is an
acceptable alternative.

A6. MORE ABOUT STATISTICS

A6.1 The Gaussian Distribution and Confidence Intervals: visualize the utility of the standard deviation, the following
A6.1.1 Often random error has a normal (Gaussian) distri- substitution can be made:
bution (16,17). This is generally true when there is no single ~x 2 µ!
y5 (A6.3)
dominant source of error. We can define a population as a σ
hypothetical set of N observations from which the sample of ~ 2y 2 !
observations actually obtained are taken. In a normally distrib- p ~ y ! dy 5 ~ 2π ! exp dy (A6.4)
2
uted population, the probability that the measured quantity has
a value between x and x + dx is given by p(x) dx as follows: The area under the distribution curve represents the prob-
ability that a measured value will be within a particular range,
p ~ x ! dx 5
~ 2π !
σ
21/2
exp F 2~x 2 µ !
2σ 2
2

G dx (A6.1) and the confidence interval placed upon a measured value can
be given in terms of the standard deviation. For example, if the
where µ is the true value of the measured quantity and σ is error is normally distributed, 68.26 % of the time the value of
the true standard deviation of the measurements. σ is defined: the single measurement will lie in the range between −σ and σ;

F G 95.44 % of the time the measurement will lie between −2σ and
n 1/2

( ~x
i51
i 2 µ !2 2σ; and 99.74 % of the time the measurement will lie between
σ5 (A6.2) −3σ and 3σ.
N

where N is the number of measurements. The standard A6.2 Gaussian Distribution of a Sample:
deviation, σ, has the same units as the quantity being measured
A6.2.1 In practice it is necessary to deal with a limited
and expresses the degree of scatter among the measured
number of measurements, n (14, 24, 25) called a sample, which
quantities and is related to the precision of the measurements.
The quantity σ2 is called the variance and has additive are obtained from the population. From these n observations,
properties. The overall variance of a multistep process is the arithmetic mean, x̄, is taken as an estimate of the true
simply the sum of the variances of the individual steps if there population mean, µ, where:
n
is no correlation between the sources of variation of the
individual steps. The square root of the overall variance yields (x
i51
i
X̄ 5 (A6.5)
the standard deviation for the whole process. To more fully n

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
13
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
A6.2.2 The operational definition of the standard deviation ~ x̄ 2 µ ! n 1/2
t5 (A6.8)
must be modified for the sample, and it is given the symbol s, s
where:
where the symbols have their usual meanings. Slight modi-

F( G
n 1/2
2 fications are necessary for the second case as follows:
~ x i 2 x̄ !
s5
i51

n21
(A6.6)
t5
~ x̄ 1 2 x̄ 2 !
s S n 1n 2
n 1 1n 2 D 1/2
(A6.9)

s2 is then the variance for the sample. The confidence ~ n 1 2 1 ! s 12 1 ~ n 2 2 1 ! s 22

interval of the mean for the sample is determined from the s2 5 (A6.10)
n 1 1n 2 2 2
Student’s t distribution as follows:
where the subscripts refer to the different data sets.
ts
µ 5 x̄6 1/2
n
(A6.7) A6.3.2 The value of t calculated from the experimental data
is compared to table values. It is clear that the value of t will
Tables of Student’s t distribution for appropriate significance become larger as the difference between x̄ and µ (or x̄1 and x̄2)
levels and appropriate n values (or degrees of freedom, n − 1) becomes larger, and large t-values imply statistically signifi-
are given elsewhere (25-28). cant differences between the quantities being compared. The
table value of t is chosen according to the significance level
A6.3 Significance Testing: required and the number of degrees of freedom (n − 1 for the
A6.3.1 The t-Test—The t-test (14, 25, 26) is useful (1) for first case and n1 + n2 − 2 for the second case). If the calculated
comparing the mean of an experimental study with an accepted value of t equals the table value for the 95 % significance level,
value or (2) for comparing the means of two sets of data. In the then only 5 % of the time would random error account for the
first case t is defined (by rearranging Eq A6.7) as follows: difference in the quantities being compared.

A7. ADVANCED STATISTICAL METHODS

A7.1 Detailed descriptions of multilinear regression (also analysis, and analyses using derivative techniques. Because of
called inverse least-squares), principal components regression the many possible variations of these methods and their relative
and partial least squares are discussed in Practices E1655. complexity, only the general nature of these methods is
Other methods of multivariate analyses are briefly discussed presented.
below as classical least squares, cross-correlation, factor

A8. CLASSICAL LEAST-SQUARES METHODS

A8.1 When the quantitative equations relating A and c are A 5 KC1E (A8.1)
written in the form of Beer’s law (see Eq 4), then the where the A matrix is an i × m matrix of m calibration
least-squares analysis is a classical regression (often referred to spectra at i wavenumbers, and C is an n × m matrix of the n
in infrared spectroscopic literature as the K-matrix method) known component concentrations for each of the m mixtures. K
(28-35). Either univariate methods (that is, each reference is the i × n matrix of the n pure component spectra at unit
spectrum contains only spectral features of one component in concentration and unit relative path length for each of the i
the spectral region of interest (28, 29, 31) or multivariate wavenumbers. E is the i × m matrix of random noise or error
methods (that is, the calibration standards are mixtures of the in the absorbance values of the calibration spectra. K is
pure components) (30) may be used. The univariate method is estimated by least-squares methods as follows:
simply a special case of the multivariate technique. When the
multivariate methods are employed, the least-squares analysis K̂ 5 AC' ~ CC' ! 21 (A8.2)
is a two-step analysis involving both a calibration step and an where K̂ is the least-squares estimate of K, the primes
analysis step. The calibration step takes the m known calibra- indicate transposed matrices and the superscript −1 indicates the
tion mixture spectra and estimates the pure component spectra inverse of the matrix. The estimated K matrix, K̂, is then used
by least-squares methods. The calibration equations are as in the least-squares prediction of the unknown sample concen-
follows: tration as follows:

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
14
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)

Ĉ 5 ~ K̂K̂' !
21
K̂' A (A8.3) independent of the number of wavenumbers used in the
where Ĉ is the least-squares estimate of the unknown sample analysis, the complexity of matrix inversion is also indepen-
concentrations and A is now the unknown sample spectrum at dent of the number of wavenumbers used. This favors the use
i wavenumbers. In the univariate case, the K matrix is simply of a large number of wavenumbers in the analysis which
constructed directly from the measured pure component spec- improves bias and precision and makes analyses possible even
tra. Baseline fitting of the sample spectrum can also be when the signal-to-noise ratios are less than one (28). Note that
included in the solution of Ĉ in Eq A8.3 to allow for baseline the use of a larger number of wavenumbers also increases the
variations between the calibration and sample spectra (28, 29). chance that the analysis will be compromised by an unknown
A weighted least-squares analysis may also be used to give component. The effect of an unknown component can be
greater emphasis to those spectral data with greater signal-to- reduced if the algorithm is written to minimize the influence of
noise ratios (29, 30). In Eq A8.2 and A8.3, the final number of those regions of the spectra that contain the unknown interfer-
simultaneous equations to be solved can be reduced to the ences (29, 30). The minimum number of calibration standards
number of components used in the analysis. This means that and wavenumbers necessary for the analysis is equal to the
the dimension of the matrix which must be inverted is also the number of components.
same as the number of components. Since this dimension is

A9. CROSS-CORRELATION METHODS

A9.1 The use of cross-correlation methods in infrared spec- mine quantitatively the amount of one contained in the other. If
troscopy has recently been described (20, 32, 33). Cross- Beer’s law is valid, then the value of the correlation function at
correlation techniques can evaluate the similarity between two zero displacement is a linear function of concentration. The
spectra (for example, the reference and the sample) to deter- mathematical details are given (20, 32, 33).

A10. ABSTRACT FACTOR METHODS

A10.1 Abstract factor methods utilize the concept of repre- Analysis of an unknown sample is then accomplished by
senting spectral data as a linear data set. Instead of using determining the scores of the principal factors for the sample
discrete data points representing changes in component spectrum. The relation between scores and concentrations
concentration, factor analysis uses the entire spectrum of each determined from the sample data can then be used to yield the
standard and transforms the data into a linear representation of sample concentrations. The linear regression analysis may be
quantitative information. Factor abstract methods involve find- carried out using any of a number of factor spaces (or basis
ing a set of principal factors (for example, eigenvectors or sets) including Fourier and Taylor series or more commonly,
loadings) required to reproduce the calibration spectra. The principal components analysis (PCA or PCR) and partial least
relation between the scores of these principal factors for each squares (PLS). Theories and application of factor analysis
calibration spectrum and the desired properties is determined methods are presented in the literature (9, 10, 21, 36) and
by the application of standard linear regression methods. Practices E1655.

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
15
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)

A11. DERIVATIVE METHODS

A11.1 Derivative methods are useful when the absorption tude of the negative-derivative lobe, or the difference between
band or bands being analyzed do not yield a suitable choice for the two lobes. The ratio of any of the above quantities between
a baseline as described in Sections 12 and 13. Derivative corresponding sample and reference bands is assumed to be
techniques have generally required that isolated spectral bands proportional to the ratio of sample and reference concentra-
be used in the analysis although one least-squares full- tions.
spectrum derivative method has been described which allows A11.1.2 Use of Second Derivatives—The second-derivative
for overlapping peaks (31). The derivative of a band or spectrum may be calculated from the first-derivative spectrum
spectrum can often be calculated by taking the difference in the same manner that the first-derivative spectrum was
between absorbances in the spectrum at equally spaced wave- calculated from the absorption spectrum. Generally the second-
numbers. However, because derivatives increase spectral noise, derivative spectrum has a positive lobe, a negative lobe, and a
mathematical smoothing (for example, Savitzky-Golay second positive lobe. The magnitude of any of these or their
smoothing) is often used either before or after derivatives are differences can be used in the quantitative analysis.
calculated.
NOTE A11.1—Depending upon the specific algorithm used, the first
A11.1.1 Use of First Derivatives—There are three possible derivative of the first derivative may not be exactly identical to the second
methods used in first-derivative analyses. These include using derivative. This is not normally a problem, as long as a consistent
the magnitude of the positive first-derivative lobe, the magni- approach is used.

A12. NOTES ON OPERATION OF NONCOMPUTERIZED, DISPERSIVE INSTRUMENTS

A12.1 For noncomputerized, dispersive instruments, the paper at the effective zero transmittance as shown in Fig.
zero value of transmittance must be known in the spectral A12.1. If the instrument does not record this point as zero
region where measurements are taken. Check the zero trans- transmittance, the value of T0 is noted and transmittance values
mittance value at the analytical wavenumber by the following are corrected by subtracting the T0 value before conversion to
procedure: absorbance. Blocking the sample beam with an opaque object
A12.1.1 When increasingly thick samples are placed in the is sometimes necessary, but gives a less accurate zero. The true
instrument, the absorption bands will bottom out on the chart absorbance A is calculated by the following equation:

FIG. A12.1 Transmittance Zero Determination

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
16
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
A 5 log@ ~ T 2 2 T 0 ! / ~ T 1 2 T 0 ! # (A12.1) T2 = transmittance without the sample in the beam.
where: If T0 = 0, then Eq A12.1 reduces to the following:
T0 = transmittance with a very thick or concentrated sample A 5 log~ T 2 /T 1 ! (A12.2)
in the beam (or with the beam blocked),
T1 = transmittance obtained with the sample in the beam,
and

A13. SETTING UP AN ANALYTICAL PROCEDURE

A13.1 Record the interval of the spectrum containing the A13.3.5 Convert these transmittance values to absorbance
band(s) to be measured. This interval should normally include and determine the analyte absorbance.
at least one maximum and one minimum transmittance.
A13.4 For computerized data, do the following:
A13.2 For each of the analytical wavenumbers that are A13.4.1 Record the spectrum of the analyte sample.
specified in the method make certain all peaks to be measured
are in the desired absorbance range. A13.4.2 Record the spectrum of a blank if possible.

A13.3 For instruments without a computer, do the following A13.4.3 Subtract the blank from the sample spectrum scal-
at the selected wavenumbers: ing appropriately.

A13.3.1 Measure the zero transmittance with a thick or A13.4.4 Using a preselected baseline treatment, determine
concentrated sample (see Annex A12). the analyte absorbance.
A13.3.2 Read the transmittance value T1 for the sample (see NOTE A13.1—For dispersive instruments, a reference beam attenuator
Fig. A12.1). is useful for shifting the apparent zero absorbance for highly absorbing
samples. For optical null instruments, changing the position of the
A13.3.3 If possible, prepare a blank that is free of analyte, attenuator will require readjustment of the gain for maximum sensitivity.
then measure T2. NOTE A13.2—Most computerized instruments have software for quan-
titative analysis which give step-by-step assistance in both calibration and
A13.3.4 Using a preselected baseline treatment, determine analysis. The general procedure is, however, as outlined in A13.1 –
the analyte and baseline transmittances (see Note 3). A13.4.4.

REFERENCES

(1) Ingle, J. D. and Crouch, S. R., Spectrochemical Analysis, Prentice (13) Cole, K. C., Thomas, Y., Pellerin, E., Dunoulin, M. M., and Paroli,
Hall, Englewood Cliffs, NJ, 1988. R. M., Applied Spectroscopy , Vol 50, No. 6, 1996, pp. 774-780.
(2) Ramsay, D. A., Journal of the American Chemical Society, Vol 74, (14) Skoog, D. A., and West, D. M., Principles of Instrumental Analysis,
1952, p. 72. 2nd edition, Saunders, Philadelphia, 1980.
(3) Anderson, R. J., and Griffiths, P. R., Analytical Chemistry, Vol 47, (15) Rothman, L. D., Crouch, S. R., and Ingle, J. D., Analytical
1975, p. 2339. Chemistry, Vol 47, 1975.
(4) Zhu, C., and Griffiths, P. R., Applied Spectroscopy, Vol 47, 1988, pp. (16) Infrared Application Study, No. 10, Perkin-Elmer Corp.
1403 and 1409. (17) Ferraro, J. R., and Basile, L. J., Fourier Transform Infrared
(5) Hieftje, G. M., Dehonigs, D., Hirschfeld, T., Applied Spectroscopy, Spectroscopy, Vol 2, Academic Press, New York, Chapter 2, 1979.
Vol 39, No. 2, p. 253. (18) Henniker, J. C., Infrared Spectrometry of Industrial Polymers,
Academic Press, 1967.
(6) Bauman, R. P., Absorption Spectroscopy, John Wiley and Sons, Inc.,
(19) Brandrup, J., and Immergut, E. H., Polymer Handbook, 2nd edition,
New York, 1962.
John Wiley and Sons, New York, 1975, p. III-51.
(7) Potts, W. J., Jr., Chemical Infrared Spectroscopy: Volume 1,
(20) Mann, C. K., Goleniewski, J. R., and Sismanidis, C.A., Applied
Techniques, John Wiley and Sons, Inc., New York, 1963. Spectroscopy, Vol 36, 1982, p. 223.
(8) Conley, R. T., Infrared Spectroscopy, 2nd edition, Allyn and Bacon, (21) Malinowski, E. R., and Howery, D. G., Factor Analysis in
Inc., Boston, 1972. Chemistry, John Wiley and Sons, Inc., New York, 1980.
(9) Kramer, R., Chemometric Techniques for Quantitative Analysis, (22) Seisler, H. W., and Holland-Moritz, K., Infrared and Raman Spec-
Marcel Decker, New York, NY, 1998. troscopy of Polymers, Vol 4, Marcel Deker, Inc., New York, 1980, p.
(10) Beebe, K. R., Pell, R. J., Seasholtz, M.B., Chemometrics, A Practical 116.
Guide, John Wiley and Sons, Inc., New York, NY, 1998. (23) An Infrared Atlas for the Coatings Industry, Federation of Societies
(11) Griffiths, P. R., and deHaseth, J. A., Fourier Transform Infrared for Coatings Technology, Philadelphia, 1980, p. 17.
Spectroscopy, John Wiley and Sons, Inc., New York, 1986. (24) Meyer, S. L., Data Analysis for Scientists and Engineers, John Wiley
(12) Smith, A. L., Applied Infrared Spectroscopy, John Wiley and Sons, and Sons, New York, 1975.
Inc., New York, 1979. (25) Fisher, R. A., The Design of Experiments, Hafner, New York, 1971.

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
17
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
 E168 − 16 (2023)
(26) Selby, S. M., ed., Standard Mathematical Tables, 15th edition, (32) Tyson, L. L., Ling, Y. C., and Mann, C. K., Applied Spectroscopy,
Chemical Rubber Co., Cleveland, 1967. Vol 38, 1984, p. 663.
(27) Fisher, R. A., and Yates, F., Statistical Tables, Hafner, Darien, CT, (33) Tyson, L. L., Vickers, T. J., and Mann, C. K., Applied Spectroscopy,
1970. Vol 38, 1984, p. 697.
(28) Haaland, D. M., and Easterling, R. G., Applied Spectroscopy, Vol 34, (34) Antoon, M. K., Keonig, J. H., and Koenig, J. L., Applied
1980, p. 539. Spectroscopy, Vol 31, 1977, p. 518.
(29) Haaland, D. M., and Easterling, R. G., Applied Spectroscopy, Vol 36, (35) Brown, C. W., Lynch, P. F., Obremski, R. J., and Lavery, D. S.,
1982, p. 665. Analytical Chemistry, Vol 54, 1982, p. 1472.
(30) Haaland, D. M., Easterling, R. G., and Vopicka, D. A., Applied (36) Mark, H., and Workman, J., Statistics in Spectroscopy, Academic
Spectroscopy, Vol 39, 1985, p. 73.
Press, Boston, 1991.
(31) Kisner, H. J., Brown, C. W., Kavarnos, G. J., Analytical Chemistry,
Vol 54, 1982, p. 1479.

ASTM International 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 International 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 International, 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 [email protected] (e-mail); or through the ASTM website
(www.astm.org). Permission rights to photocopy the standard may also be secured from the Copyright Clearance Center, 222
Rosewood Drive, Danvers, MA 01923, Tel: (978) 646-2600; http://www.copyright.com/

&RS\ULJKWE\$670,QW
ODOOULJKWVUHVHUYHG7KX)HE*07
18
'RZQORDGHGSULQWHGE\
2UWD'RJX7HNQLN8QLYHUVLWHVLSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG