tC·-s Readout Devices

A readout device is a transducer that converts information from

_an electrical domain to a form that is understandable by a human
observer. Usually, the transduced signal takes the form of the
alphanumeric or graphical output of a cathode-ray tube, a series of
numbers on a digital display, the position of a pointer on a meter
scale, or occasionally, the blackening of a photographic plate or a
tracing on a recorder paper. In some instances, the readout device
may be arranged to give the analyte concentration directly.

1(-6 Computers in Instruments

Most modern analytical instruments contain or are attached to
one or more sophisticated electronic devices and data-domain
converters, such as operational amplifiers, integrated circuits,
analog-to-digital and digital-to-analog converters, counters,
microprocessors, and computers. To appreciate the power and
limitations of such instruments, investigators need to develop at
least a qualitative understanding of how these devices function
and what they can do. Chapters 3 and 4 provide a brief treatment
of these important topics.

1D CALIBRATION OF INSTRUMENTAL
METHODS
A very important part of all analytical procedures is the calibra-
tion and standardization process. Calibration determines the
relationship between the analytical response and the analyte
concentration. Usually this is determined by the use of chemical
standards.
10 Chapter 1 Introduction (((

. [Link] all analytical methods require some type of calibration A calibration curve is prepared by plotting the data or by fitting
with chemical standards. Gravimetric methods and some coulo- them to a suitable mathematical equation , such as the slope.
metric methods (Chapter 24) are among the few absol11te methods intercept form used in the method of linear least squares. The
that do not rely on calibration with chemical standards. Several next step is the prediction step, where the response signal is
types of calibration procedures are described in this section. obtained for the sample and used to predict the unknown analyte
concentration . ' x• from the calibration curve or best-fit equation.
The con cen tration of the analyte in the original bulk sample
1D-1 Comparison with Standards is then calculated from cx by applying the appropriate dilution
Two types of comparison methods are described here, the direct factors from the sample preparation steps.
comparison technique and the titration procedure.
The Least-Squares Method
Direct Comparison A typical ca(jbration curve is shown in Figure 1-8 for the deter-
Some analytical procedures involve comparing a property of mination of Ni(II) by reaction with excess thiocyanate to form
the analyte (or the product of a reaction with the analyte) with an absorbing complex [Ni(SCN)"' ]. The absorbance of the com-
standards such that the property being tested matches or nearly plex is measured for a series of Ni(II) [Link] solutions and
matches that of the standard. For example, in early colorime- plotted as a function of Ni(II) concentration. The ordinate is the
ters, the color produced as the result of a chemical reaction of dependent variable, absorbc1nce, and the absdssa is the indepen-
the analyte was compared with the color produced by reaction dent variable, concentration of Ni(II). As is typical and usually
of standards. If the concentration of the standard was varied desirable, the plot approximates a straight line. Note, however,
by dilution, for example, it was possible to obtain a fairly exact that because of the indeterminate errors in the measurement
color match. The concentration of the analyte was then equal to process, not all the data fall exactly on the line. Thus, th.e inves-
the concentration of the standard after dilution. Such a proce- tigator must try to draw the "best" straight line among the data
dure is called a null comparison or isomation method.
3
points. Regression analysis provides the means for objectively
obtaining such a line and also for specifying the uncertainties
Tit rations associated with its subsequent use. The uncertainties are related
Titrations are among the most accurate of all analytical pro- to the residuals shown in Figure 1-8, which are a measure of
cedures. In a titration, the analyte reacts with a standardized how far away from the best straight line the data points lie. The
reagent (the titrant) in a reaction of known stoichiometry. Usu- method of least squares (see Appendix 1, Section alD) is often
5
ally the amount of titrant is varied until chemical equivalence is applied to obtain the equation for the line.
reached, as indicated by the color change of a chemical indicator The method of least squares is based on two assumptions.
or by the change in an instrument response. The amount of the
The first is that there is actually a linear relationship between the
standardized reagent needed to achieve chemical equivalence
measured response y and the standard analyte concentration x.
can then be related to the amount of analyte present. The titra-
The mathematical relationship that describes this assumption is
tion is thus a type of chemical comparison.4 called the regression model, which may be represented as
y=mx+b
1D-2 External-Standard Calibration
An external standllrd is prepared separately from the sample. By where bis they intercept (the value of y when xis zero) and m
contrast, an internal standard is added to the sample itself.. Exter- is the slope of the line (see Figure 1-8). We also assume that any
nal standards are used to calibrate instruments and procedures deviation of the individual points from the straight line arises
when there are no interference effects from matrix components from error in the measurement. That is, we assume there is no
in the analyte solution. A series of such external standards con- error in the x values of the points (concentrations). Both of these
taining the analyte in known concentra~ions is prepared. Ideally, assumptions are appropriate for many analytical methods, but
three or more such solutions are used in the calibration process. bear in mind that whenever there is significant uncertainty in
However, in some routine analyses, two-point calibrations can the x data, basic linear least-squares analysis may not give the
best straight line. In such a case, a more complex correlatioll
be reliable.
analysis may be necessary. In addition , basic !east-squares
Calib ration is accomplished by obtaining the response
signal (absorbance, emission intensity, electrode potential,

•
peak area) as a function of the known analyte concentration.
Tutorial: Learn more about calibration at
[Link]/ skoogpia 7.
' Sec , for oamplc, H. V. Malnutadt and J. D. Wincfordncr, Anal. Chim . Acta,
19S9. 20, 283 . DOI : 10.1016/0003 -2670(59)80066•0; L Ramaley and C. G. Enke,
Anal O,,m , 1965. 37, 1073. DOI; I0.1021 /ac60227a041. ' For a discussion of using spreadsheets in Uncar regression analysis. see~-~-
Crouch and I'. J. Holler, Applications of Microsoft• Excel;,, Am1/ytfcal ChrmutrJ·,
•~« D ,., Skoog. D. M. Wut, I'. J. Holler. [Link] S. R. Crouch , Fundamttrtals of
M,alyli, al Cht miw y, 9\h ed.. Sc:lmont, CA: Brooks/Cole, 2014., Chaps. 13- 17. 3rd w., 8elmool, CA: Cengage Learning, 20 17, Chop. 4.
 ))) 10 Calibration of Instrumental Methods 11

Residual = y1 - (ttu 1 + b)

0.9

0.8 I-

0.7

0.6
I
.D I
6 0.5 -- -----,---- I
V: - t
--
.D
<
.,:_ 0.4 - · - -
--- '
0.3 i-- . ~- L --
0.2 r - ' -:--

0 '-----'--,---'---'--
--1---.-;- 1 :

-I

' - - - - '...'--- - - - -- - - - - - - - '

I
--+--
0.000 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000
x, Concentration ofNi(m X 103, M
FIGURE 1-8 Calibration curve for the determination of Ni(II) by formation of an absorbing
species. The absorbance of standards are shown as solid circles. The distance on the y-axis
between a data point and the predicted line is called a residual as shown in the inset. The
calibration curve is used in an inverse fashion to find the concentration of an unknown.

analysis may not be appropriate when the uncertainties in they for the unknown, and the slope and intercept are used to cal-
values vary significantly with x. In this case, it may be necessary culate the unknown concentration '" as shown in Equation 1-1.
to apply different weighting factors to the points and perform a
y, - b
weighted [Link]-squares analysis.6 ex=-;;;- (1-1)
In cases where the data do not fit a linear model, nonlinear
regression methods are available. 7 Some of these use polynomial The standard deviation in concentration s, can be found
models or multiple regression procedures. There are even com- from the standard error of the estimates,. also called the standard
puter programs that will find a model that describes a set of exper- deviation about regression, as given in Equation 1-2,
imental° data from an internal or user-defined set of equations.
Sy 1 1 (y, - y)2
The slope m and intercept b of the linear least-squares line s =- -+-+-'-----=--
2
(1-2)
' m M N m S""
are determined as in Equations al-34 and al -35 of Appendix 1.
For determining an unknown concentration'" from the least- where M is the number of replicate results, N is the number of
squares line, the value of the instrument response y, is obtained points in the calibration curve (number of standards), y, is the
mean response for the unknown, and y is the mean value of y for
the calibration results. The quantity S"" is the sum of the squares
"see P. R. Bevington and D. K. Robinson, Data Reduction and Error Analysis of the deviations of x values from the mean as given in Equation
for the Physical Sciences, 3rd ed., New York: McGraw-Hill, 2003; for an Excel al-31 of Appendix 1.
approach to weighted .linear regression, see S. R. Crouch and F. J. Holler, Applica-
tions of Microsoft" Exal in Analytical Chemistry, 3rd ed., Belmont, CA: Cengage
Learning. 2017. pp. 331-337.
Errors in External-Standard Calibration
7
J. L. Devore, Probability and Statistics for Engineering and the Sciences, 9th ed., When external standards are used, it is assumed that the same
Pacific Grove, Ck Duxbury Press at Brooks/Cole, 2016. responses will be obtained when the same analyte concentration
 12 Chapt, r l Introduction <«

10

2.5

3 5 7 9
Concentration
FIGURE 1-9 Effect of calibration curve uncertainty. The dashed lines show confidence limits
for concentrations determined from the reg ression line. Note that uncertainties increase at
the extremities of the plot. Usually, we estimate the uncertainty in analyte concentration
only from the standard deviation of the response. Calibration curve uncertainty can
significantly increase the uncertainty in analyte concentration from s, to s~.

is present in the sample and in the standard. Thus, the calibra- Also, systematic errors can occur during the calibration
tion functional relationship between the response and the ana- process. For example, if the standards are prepared incorrectly,
lyte concentration must apply to the sample as well. Usually, in an error will occur. The accuracy with which the standards are
a determination, the raw response from the instrument is not prepared depends on the accuracy of the gravimetric and volu-
used. Instead, the raw analytical response is corrected by mea- metric techniques and equipment used. The chemical form of
suring a blank. An ideal blank is identical to the sample but the standards must be identical to that of the analyte in the sam-
without the analyte. In [Link], with complex samples, it is too ple; the state of oxidation, isomerization, or complexation of the
time-consuming or impossible to prepare an ideal blank and a analyte can alter the response. Once prepared, the concentration
compromise must be made. Most often a real blank is either a of the standards can change because of decomposition, volatil-
solvent blank, containing the same solvent in which the sample is ization, or adsorption onto container walls. Contamination of
dissolved, or a reagent blank, containing the solvent plus all the the standards can also result in higher analyte concentrations
reagents used in sample preparation. than expected. A systematic error can occur if there is some bias
Even with blank corrections, several factors can cause the in the calibration model. For example, errors can occur if the
basic assumption of the external-standard method to break calibration function is obtained without using enough standards
down. Matrix effects, due to extraneous species in the sample to obtain good statistical estimates of the parameters.
that are not present in the standards or blank, can cause the Random errors can also influence the accuracy of results
same analyte concentrations in the sample and standards to obtained from calibration curves, as illustrated in Figure 1-9.
give different responses. 8 Differences in experimental variables The uncertainty in the concentration of analyte s: obtained
at the times at which blank, sample, and standard are mea- from a calibration curve is lowest when the response is close to
sured can also invalidate the established calibration function . the mean value y. The point x,y represents the centroid of the
Even when the basic assumption is valid, errors can still occur regression line. Note that measurements made near the center of
because of contamjnation during the sampling or sample prepa- the curve will give less uncertainty in analyte concentration than
[Link] steps. those made at_the extremes.

Multivariate Calibration
'Tht matrix includt1 the analytt and oth t r constil ucnt s. wh ich arc termed The least -squares procedure just described Is an example of 0
tont.'omitanu. univariate calibration procedure because only one response Is
 »> 10 Calibration of Instrumental Methods 13

used per sample. The process of relating multiple instrument where k is a proportionality constant. A plot of Sas a function of
responses to an analyte or a mixture of analyles is known as V, is a straight line of the form
mu ltivariate ca lib ration . Multivariate calibration methods 9
have become quite popular In recent years as new instruments S = mV, + b
become available that produce multidimensional responses where the slope m and the intercept b are given by
(absorbance of several samples at multiple wavelengths, mass
spectrum of chromatographically separated components, etc.). kc,
m= -
Multivariate calibration methods are very powerful. They can be V1
used to simultaneously determine multiple components in mix- and
tures and can provide redundancy in measurements to improve
precision because repeating a measurement N times provides b = kVxcx
a VN improvement in the precision of the mean value (see V,
Appendix l, Section alB-1). They can also be used to detect the
Such a standard-addition plot is shown in Figure 1-1O.
presence of interferences that would not be identified in a uni-
variate calibration. A least-squares analysis (Appendix I, Section alD) can be
used to determine m and b; ex can then be obtai ned fro m the
ratio of these two quantities and the known values of c,, V.., and
1D-3 Standard-Addition Methods V,. Thus,
Standard-addition methods are particularly useful for analyzing
complex samples in which the likelihood of matrix effects is sub-
m kc,/V 1
stantial. A standard-addition method can take several forms. 10
One of the most common forms involves adding one or more
'·
or
increments of a standard solution to sample aliquots containing
identical volumes. This process is often called spiking the sample. be,
C =-- (1-4)
Each solution is then diluted to a fixed volume before measure- z mVx
ment. Note that when the amount of sample is limited, stan-
dard additions can be carried out by successive introductions of The standard deviation in concentration can then be
increments of the standard to a single measured volume of the obtained by first calculating the standard deviation in volume
unknown. Measurements are made on the original sample and on Sy and then using the relationship between volume and

the sample plus the standard after each addition. In most versions concentration. The standard deviation in volume is found from
of the standard-addition method, the sample matrix is nearly Equation 1-2 with a few alterations. Because we extrapolate the
identical after each addition, the only difference being the con- calibration curve to the x-axis in the standard-addition method,
centration of the analyte or, in cases involving the addition of an the value of y for the unknown is Oand the 1/M term is absent.
excess of an analytical reagent, the concentration of the reagent. Hence, the equation for sv becomes
All other constituents of the reaction mixture should be identical Sy l (0 - ji)2
because the standards are prepared in aliquots of the sample. Sy= - - + -2- - (1-5)
m N m S.a
Assume that several aliquots Vz of the unknown solution
with a concentration ex are transferred to volumetric flasks As shown by the dashed line of Figure 1-10, the difference
having a volume Vt. To each of these flasks is added a variable between the volume of the standard added at the origin (zero)
volume V, of a standard solution of the analyte having aknown and the value of the volume at the intersection of the straight
concentration , .. Suitable reagents are then added, and each line with the x-axis, or the x-intercept (V,) 0, is the volume of
solution is diluted to volume. Instrumental measurements are standard reagent equivalent to the amount of analyte in the sam-
then made on each of these solutions and corrected for any ple. In addition, the x-intercept corresponds to zero instru~ent
blank response to yield a net instrument response S. If the blank- response, so that we may write
corrected instrument response is proportional to concentration, kV,c, kVxcx
as is assumed in the standard-addition method, we may write S=--+--=0 ( 1-6)
V1 Vt

(1-3) By solving Equation 1-6 for ex, we obtain

(V,) 0c,
(1-7)
~For a more nlcnsive discussion, sec K. Varmuza and P. Filzmoscr, Introduction
C
x= - -vx-
to Multivariate Statistical Analyiis in Chemometrlci, Boca Raton, FL: CRC Press,
2009: K. R. Beebe, R. J. Pell, and M. B. Scasholtz, Chemometrics: A Practical Guide, The standard deviation in concentration s, is then
New York: Wiley, 1998. Chap. 5; H. Marten• and T. Naes, Multivariate Calibration,
New York: Wiley, 1989. (1-8)
'"Sec M. Bad~./. Chem. Educ., 1980, 57, 703, DOI: 10.1021/ed057p703.
14 Chapter 1 Introduction «<

1.2

1.0
Ill = 0.03820
0.8
"u
C

"
0 0.6

!/
"' ( V,Jo = - 6.314 mL
.,; (calculated or extrapolated)
0.4

0.2
,,.
,,.
0.0 10.0 20.0
-10.0 0.0
V,, mL

FIGURE 1-10 Linear calibration plot for the method of standard additions. T~e concen-
tration of the unknown solution may be calculated from the slope m and the intercept
b, or it may be determined by extrapolation, as explained in the text.

Example 1-1 shows the uses of these quantities. is -6.314 mL. The unknown concentration of the ana-
lyte in the original solution is then calculated from
Equation 1-7 as follows:
EXAMPLE 1-1 ( V,) 0c, 6.314 mL X 11.1 ppm
C = - --=
X V, 10.00mL
Ten-millimeter aliquots of a natural water sample were pipet-
ted into 50.00-mL volumetric flasks. Exactly 0.00, 5,00, 10.00, = 7.01 ppm FeH
15.00, and 20.00 mL of a standard solution containing 11.1 ppm
(b) The standard deviation in concentration can be obtained
of Fe 3 + were added to each, followed by an excess of thiocya ·
from the standard deviation in the volume inter-
nate ion to give the red complex Fe(SCN)H. After dilution to
cept (Equation 1-5) and Equation 1-8 as shown in the
volume, the instrument response S for each of the five solu-
spreadsheet of Figure 1-11. The result is s, = 0.16 ppm.
tions, measured with a colorimeter, was found to be 0.240,
Hence, the concentration of Fe in the unknown is
0.437, 0.621, 0.809, and 1.009, respectively. (a) What was the
7.01 ± 0.16 ppm.
concentration of FeH in the water sample? (b) Calculate the
standard deviation in the concentration of Fe 3 + .
In the interest of saving time or sample, it is possible to per-
Solution form a standard-addition analysis by using only two increments
(a) In this problem, c, = 11.1 ppm, V, = [Link] mL, and of sample. Here, a single addition of V, mL of standard would be
V, = 50.00 ml. A plot of the data, shown in Figure 1-10, added to one of the two samples, and we can write
demonstrates that there is a linear relationship between
the solution absorbance and the added amount of iron.
To obtain the equation for the line in Figure 1-10 (S =
mV, + b), we follow the procedure illustrated in Example kVxc, kV,c,
a1-11 in Appendix 1. The result, shown in the spreadsheet in S2=--+--
V, V,
Figurel-11, ism = 0.0382andb = 0.2412andthus
where S1 and S2 are the signals resulting from the diluted sample
S = 0.0382V, + 0.2412 and the diluted sample plus standard, respectively. Dividing the
From Equation 1-4, or from the spreadsheet, we obtain second equation by the first gives, after rearranging,

c = 0.2412 X 11.1 = 7.01 ppm FeH

' 0.0382 X 10.00
This value may be determined by graphical extrapola- The single-addition method is somewhat dangero us because
tion as the figure also illustrates. The extrapolated value it presumes a linear relationship and provides no check ofthls
represents the volume of reagent corresponding to zero assumption. The multiple-addition method at least gives a check
instrument response (absorbance), which in this case on the linearity supposition.

F
 )}) 10 Calibration of Instrumental Methods 15

A B I C I D I E I F
1 Determination of Fe In Natural Water by Colorlmoll'l with Multiple Additions I G I H I I J
2 Concentration of standard , c9 11 .10 ppm I I I J
3 Volume of unknown used, V, 10.00 ml 1.2
4 Volume of standard added Absorbance, A
5 0.00 0.240
6 5.00 0.437
7 10.00 0.621 1.0 y = 0.0382x + 0.2412
1
8 15.00 ' 0.809 R = 0.9998
9 20.00 1.009
10
11 Re11ra11lon Equation
0.8
12 Slope
0.0382
13 Intercept
0.2412
14 Volume interc~t
-6.31414
15 Concentration of unknown
7.01 IPDm 0.6
16 Error Analysts
17 Standard error in _y
0.004858
18 N
5
19 SJOC 0.4
250
20 IY bar
0.6232
21 Standard deviation in volume
0.143011
22 Standard deviation in c
0.16
23 Spreadsheet Documentation o;i
24 Cell B12=SLOPE(B5:B9,A5:A9
25 Cell B13=1NTERCEPT!B5:B9,A5:A9)
26 Cell B14=-B13/B12
nn
27 Cell B15=-B 14.B2183
28 Cell B17-STEYX1B5:B9,A5:A9) -10.00 -5.00 0.00 5.00 10.00 15.00 20.00 25.00
29 Cell B18=COUNT_(B5:B9) Volume of standard solution, ml
30 Cell B19=DEVSQ(A5:A9)
31 Cell B20=AVERAGE(B5:B9)
I
32 Cell B21...,(B17/B12tSORT(1/B18+((0-B20jA2)/(/812A2 *B19))
33 Cell B22=B21 *B2/B3 I
I
FIGURE 1-11 Spreadsheet for standard-addition Example 1-1.

10-4 The Internal-Standard Method samples and standards in a reproducible way. The internal stan-
An internal standard is a substance that is added in a constant dard should provide a signal that is similar to the analyte signal
amount to all samples, blanks, and calibration standards in an in most ways but sufficiently different so that the two signals are
analysis. Alternatively, it may be a major constituent of sam- distinguishable by the instrument. The internal standard must
ples and standards that is present in a large enough amount be known to be absent from the sample matrix so that the only
that its concentration can be assumed to be the same in all source of the standard is the added amount. For example, lith-
cases. Calibration then involves plotting the ratio of the ana- ium is a good internal standard for the determination of sodium
lyte signal to the internal-standard signal as a function of the or potassium in blood serum because the chemical behavior of
analyte concentration of the standards. This ratio for the sam- lithium is similar to both analytes, but it does n~t occur natu-
ples is then used to obtain their analyte concentrations from a rally in blood.
calibration curve.
An example of the determination of sodium in blood by
An internal standard, if properly chosen and used, can com- flame spectrometry using lithium as an internal standard is
pensate for several types of both random and systematic errors. shown in Figure 1-12. The upper figure shows the normal cal-
Thus, if the analyte and internal-standard signals respond pro- ibration curve of sodium intensity versus sodium concentration
portionally to random instrumental and method fluctuations, in ppm. Although a fairly linear plot is obtained, quite a bit of
the ratio of these signals is independent of such fluctuations. If scatter is observed. The lower plot shows the intensity ratio of
the two signals are influenced in the same way by matrix effects, sodium to lithium plotted against the sodium concentration in
compensation of these effects also occurs. In those instances ppm. Note the improvement in the calibration curve when the
where the internal standard is a major constituent of samples internal standard is used.
and standards, compensation for errors that arise in sample
preparation, solution, and cleanup may also occur. In the development of any new internal-standard method,
we must verify that changes in concentration of analyte do not
A majo r difficulty in applying the internal-standard affect the signal intensity that results from the internal standard
method Is that of finding a suitable substance to serve as the
and that the internal standard does not suppress or enhance the
internal standard and of introducing that substance into both analyte signal.