Calibration Methods

A Historic Calibration Curve

10 1

10 2

[F ] in test solution

10 3

10 4

10 5

10 6 20 60 100 140 180 220 Electrode response (mV) 260 300

Laboratory notebook page of Martin Frant, showing the response of the very first fluoride electrode that he assembled at Orion Research in Cambridge, Massachusetts, in 1966. The theoretical response of the electrode is a straight line with a slope of 59 mV per decade change in F concentration. The observed response of this first electrode was 56 mV per decade. Chapter 15 describes how this electrode works.
[From M. S. Frant, Where Did Ion Selective Electrodes Come From? J. Chem. Ed. 1997, 74, 159.]

In the 1960s, there was great need for a simple, selective method to measure F concentration. Fluoride was being added for the first time (at a level of 1 ppm 1 mg/L) to municipal water supplies to prevent tooth decay. The existing analytical method for F required a tedious distillation (to separate analyte from interfering species) followed by a fickle spectrophotometric analysis. When Martin Frant noticed an advertisement for rare-earth fluoride single crystals for the newly invented laser, he recognized that these crystals could have the specific response to F for which he had been searching. The very first crystal of NdF3 that he sealed into a plastic tube with sealing wax gave the response shown above. More than 300 000 fluoride electrodes have been sold by now, and they are widely used to monitor and control F addition to the water that most of us drink.

For most chemical analyses, the response of the procedure must be evaluated for known quantities of analyte (called standards) so that the response to an unknown quantity can be interpreted. For this purpose, we commonly prepare a calibration curve, which is a

80

graph such as Figure 0-7 or the graph on the facing page showing the analytical response as a function of the known quantity of analyte present. For example, we can see that if the electrode response for an unknown is 180 mV, then the F concentration is 2.0 10 4 M. Most often, we work in a region where the calibration curve is a straight line. We use the method of least squares, described in this chapter, to draw the best straight line through experimental data points that have some scatter and do not lie perfectly on a straight line. We will learn to estimate the uncertainty in a chemical analysis based on the uncertainties in the calibration curve and in the measured response to replicate samples of unknown. We also discuss standard addition and internal standards, which are important calibration methods.

81
5-1 Finding the Best Straight Line

Standard addition and internal standards could be put off until they are required in the lab or in later chapters.

5-1 Finding the Best Straight Line

What is the best way to draw a straight line through a set of experimental points that contain some error and do not lie exactly on a straight line? In Figure 5-1, for example, the best line will be such that some of the points lie above and some lie below the line.
5

Vertical deviation = yi y

(xi, yi) y y x Slope = y =m x

3 y

Figure 5-1

y = mx + b 2

Intercept = b

3 x

Least-squares curve fitting. The points (1,2) and (6,5) do not fall exactly on the solid line, but they are too close to the line to show their deviations. The Gaussian curve drawn over the point (3,3) is a schematic indication of the fact that each value of yi is normally distributed about the straight line. That is, the most probable value of y will fall on the line, but there is a finite probability of measuring y some distance from the line.

Method of Least Squares

The method of least squares is the most widely used technique for finding a line (or a curve) through a set of data points.1 The procedure we will use assumes that the errors in the y values are substantially greater than the errors in the x values.2 This condition is usually true in a calibration curve in which the experimental response (y values) is less certain than the quantity of analyte (x values). A second assumption is that the uncertainties (standard deviations) in all the y values are similar. Suppose we seek to draw the best straight line through the points in Figure 5-1 by minimizing the vertical deviations between the points and the line. We minimize only the vertical deviations because we assume that uncertainties in the y values are much greater than uncertainties in the x values. Let the equation of the line be

Equation for a straight line: y Slope (m) y-Intercept (b) y x y2 x2 y1 x1

mx

crossing point on y-axis

y (x1, y1) b

(x2, y2) y = y2 y1 x = x2 x1 x

Equation of straight line:

mx

(5-1)

in which m is the slope and b is the intercept on the y-axis. The vertical deviation for the point (xi, yi) in Figure 5-1 is yi y, where y is the ordinate of the straight line when x xi.

82
5 Calibration Methods

Vertical deviation

di

yi

yi

(mxi

b)

(5-2)

Some of the deviations are positive and some are negative. Because we wish to minimize the magnitude of the deviations irrespective of their signs, we square all the deviations so that we are dealing only with positive numbers:

d2 i

(yi

y) 2

(yi

mxi

b) 2

To evaluate the determinant, multiply the diagonal elements e h and then subtract the product of the other diagonal elements f g.

Because we minimize the squares of the deviations, this is called the method of least squares. It can be shown that minimizing the squares of the deviations (rather than simply their magnitudes) corresponds to assuming that the set of y values is the most probable set. Finding values of m and b that minimize the sum of the squares of the vertical deviations involves some calculus, which we omit. We will express the final solution for slope and intercept in terms of determinants, which summarize certain e f arithmetic operations. The determinant 2 fg. So, for 2 represents the value eh g h example, The slope and the intercept of the best straight line are found to be

Translation of least-squares equations: m b n (xiyi ) n (x 2) i (x 2) yi i n (x 2) i xi yi ( xi ) 2 (xiyi ) xi ( xi ) 2

Least-squares best line

where D is

6 4

5 2 3

(6

3)

(5

4)

Slope: m b

Intercept:
D 2

(x2 ) i xi

(xiyi ) yi (x2 ) i 2 xi xi 2 n

xi 2 D n (xiyi ) 2 D yi

(5-3) (5-4)

(5-5)

A spreadsheet for this gruesome arithmetic is given in Figure 5-9. Scientific calculators find the slope and intercept for you but generally do not give the uncertainties.

and n is the number of points. Lets use these equations to find the slope and intercept of the best straight line through the four points in Figure 5-1. The work is set out in Table 5-1. Noting that n 4 and putting the various sums into the determinants in Equations 5-3, 5-4, and 5-5 gives

m b

The equation of the best straight line through the points in Figure 5-1 is therefore

62 14

57 14

57 2 14

14 2 4

62 14

62 14

14 2 4 y

14 2 4

(57 (62 (62 (62

4) 4) 14) 4)

(14 (14 (57 (14 1.346 15

14) 14) 14) 14)

32 52 70 52

0.615 38 1.346 15

0.615 38x

We tackle the question of significant figures for m and b in the next section.

Table 5-1 Calculations for least-squares analysis

xi 1 3 4 6 xi 14 yi yi 2 3 4 5 14 (xi yi) xi yi 2 9 16 30 57 (x 2) i x2 i 1 9 16 36 62 di ( yi mxi b) d2 i 0.001 479 3 0.036 982 0.036 982 0.001 479 3 (d 2) i 0.076 923 0.038 46 0.192 31 0.192 31 0.038 46

Example

Finding Slope and Intercept with a Spreadsheet

83
5-1 Finding the Best Straight Line

Your scientific calculator has a procedure for computing the slope and intercept of a set of (x, y) data, and you should know how to use that procedure. Alternatively, Excel has functions called SLOPE and INTERCEPT whose use is illustrated below:
A B x y 1 2 3 3 4 4 6 5 C D slope 0.61538 intercept 1.34615 E Formulas: D3 D5 F

1 2 3 4 5

SLOPE(B2:B5,A2:A5) INTERCEPT(B2:B5,A2:A5)

The slope in cell D3 is computed with the formula SLOPE(B2:B5,A2:A5), where B2:B5 is the range containing the y values and A2:A5 is the range containing x values.

How Reliable Are Least-Squares Parameters?

To estimate the uncertainties (expressed as standard deviations) in the slope and intercept, an uncertainty analysis must be performed on Equations 5-3 and 5-4. Because the uncertainties in m and b are related to the uncertainty in measuring each value of y, we first estimate the standard deviation that describes the population of y values. This standard deviation, y, characterizes the little Gaussian curve inscribed in Figure 5-1. We estimate y, the population standard deviation of all y values, by calculating sy, the standard deviation, for the four measured values of y. The deviation of each value of yi from the center of its Gaussian curve is just di yi y yi (mxi b) (Equation 5-2). The standard deviation of these vertical deviations is
y

sy

2 a (di d) B (degrees of freedom)

(5-6)

Equation 5-6 is an analogous to Equation 4-2.

Because the average deviation, d , is 0 for the best straight line, the numerator of Equation 5-6 reduces to (d 2). i The degrees of freedom is the number of independent pieces of information available. For n data points, there are n degrees of freedom. If you were calculating the standard deviation of n points, you would first find the average to use in Equation 4-2. This leaves n 1 degrees of freedom in Equation 4-2 because only n 1 pieces of information are available in addition to the average. If you know n 1 values and you also know their average, then the nth value is fixed and you can calculate it. How does this discussion apply to Equation 5-6? We began with n points. But two degrees of freedom were lost in determining the slope and the intercept of the best line. Therefore, n 2 degrees of freedom remain. Equation 5-6 becomes

If you know x and n 1 of the individual values, you can calculate the nth value. Therefore, the problem has just n 1 degrees of freedom once x is known.

sy

a (d i ) B n 2
2

(5-7)

where di is given by Equation 5-2. Uncertainty analysis for Equations 5-3 and 5-4 leads to the following results: Standard 2 sm deviation of slope and s2 intercept b

s2n y D s2 a (x2 ) y i D

(5-8) (5-9)

where sm is an estimate of the standard deviation of the slope, sb is an estimate of the standard deviation of the intercept, sy is given by Equation 5-7, and D is given by Equation 5-5.

84
5 Calibration Methods

At last, we can assign significant figures to the slope and the intercept in Figure 5-1. In Table 5-1, we see that (d 2) 0.076 923. Putting this number into Equation 5-7 i gives

s2 y

0.076 923 4 2

0.038 462

Now we can plug numbers into Equations 5-8 and 5-9 to find

s2 m s2 b

s2n y D D

(0.038 462)(4) 52 (0.038 462)(62) 52 0.615 38 0.054 39 1.346 15 0.214 15

0.002 958 6 1 sm 0.045 859 1

0.054 39 sb 0.214 15

s2 a (x2 ) i y

Combining the results for m, sm, b, and sb, we write

The first digit of the uncertainty is the last significant figure. We often retain extra, insignificant digits to avoid roundoff errors in further calculations.

Slope: Intercept:

0.62 1.3

0.05 or 0.615 0.2 or 1.35

0.054

(5-10) (5-11)

0.21

For example, the 95% confidence interval for the slope is tsm (4.303)(0.054) 0.23 based on degrees of freedom n 2 2. (The confidence interval is tsm , not tsm > 2n , because 2n is already implicit in sm.)

where the uncertainties represent one standard deviation. The first decimal place of the standard deviation is the last significant figure of the slope or intercept. If you want to express the uncertainty as a confidence interval instead of one standard deviation, multiply the uncertainties in Equations 5-10 and 5-11 by the appropriate value of Students t from Table 4-2 for n 2 degrees of freedom.

Example

Finding sy, sm, and sb with a Spreadsheet

The Excel function LINEST returns the slope and intercept and their uncertainties in a table (a matrix). As an example, enter x and y values in columns A and B. Then highlight the 3-row 2-column region E3:F5 with your mouse. This block of cells is selected to contain the output of the LINEST function. Under the INSERT menu, select FUNCTION. In the window that appears, go to Statistical and double click on LINEST. The new window asks for four inputs to the function. For y values, enter B2:B5. Then enter A2:A5 for x values. The next two entries are both TRUE. The first TRUE tells Excel that we want to compute the y-intercept of the least-squares line and not force the intercept to be 0. The second TRUE tells Excel to print out the standard deviations as well as the slope and intercept. The formula you have just entered is LINEST(B2:B5,A2:A5,TRUE,TRUE). Click OK and the slope appears in cell E3.
A 1 2 3 4 5 6 7 8 9 10 x 1 3 4 6 y 2 3 4 5 B C E F Output from LINEST Slope Intercept Parameter 0.61538 1.34615 Std Dev 0.05439 0.21414 R^2 0.98462 0.19612 D G

Std Dev (y)

Highlight cells E3:F5 Type " LINEST(B2:B5,A2:A5,TRUE,TRUE)" Press CTRL SHIFT ENTER (on PC) Press COMMAND RETURN (on Mac)

The output of LINEST should be a matrix, not a single number. What went wrong? To tell the computer that you want a matrix, go back and highlight cells E3:F5. LINEST(B2:B5,A2:A5,TRUE,TRUE) appears once again in the formula line. Now

press CONTROL SHIFT ENTER on a PC or COMMAND( ) RETURN on a Mac. Excel dutifully prints out a matrix in cells E3:F5. Write labels around the block to indicate what is in each cell. The slope and intercept are on the top line. The second line contains sm and sb. Cell F5 contains sy and cell E5 contains a quantity called R2, which is defined in Equation 29-1 and is a measure of the goodness of fit of the data to the line. The closer R2 is to unity, the better the fit.

85
5-2 Calibration Curves

5-2 Calibration Curves

A calibration curve shows the response of an analytical method to known quantities of analyte.3 Table 5-2 gives real data from a protein analysis that produces a colored product. An instrument called a spectrophotometer measures the absorbance of light, which is proportional to the quantity of protein analyzed. Solutions containing known concentrations of analyte are called standard solutions. Solutions containing all the reagents and solvents used in the analysis, but no deliberately added analyte, are called blank solutions. Blanks measure the response of the analytical procedure to impurities or interfering species in the reagents. Scanning across the three absorbance values in each row of Table 5-2, the number 0.392 seems out of line: It is inconsistent with the other values for 15.0 g, and the range of values for the 15.0- g samples is much bigger than the range for the other samples. The linear relation between the average values of absorbance up to the 20.0g sample also indicates that the value 0.392 is in error (Figure 5-2). We therefore chose to omit the value 0.392 from all subsequent calculations. It is reasonable to ask whether all three absorbances for the 25.0- g samples are low for some unknown reason, because this point falls below the straight line in Figure 5-2. Many repetitions of this analysis show that the 25.0- g point is consistently below the straight line and there is nothing wrong with the data in Table 5-2.
Sections 18-1 and 18-2 discuss absorption of light and define the term absorbance. We will use concepts from these two sections throughout this book. You may want to read these sections at this time for background.

Questionable data point 0.400 Absorbance 0.300 0.200 0.100 Average with questionable point Average without questionable point

Table 5-2 Spectrophotometer data used to construct calibration curve

Amount of protein ( g) 0 5.0 10.0 15.0 20.0 25.0 Absorbance of independent samples 0.099 0.185 0.282 0.345 0.425 0.483 0.099 0.187 0.272 0.347 0.425 0.488 0.100 0.188 0.272 (0.392) 0.430 0.496 Range 0.001 0.003 0.010 0.047 0.005 0.013 Corrected absorbance 0.0003 0.0857 0.1827 0.2457 0.3257 0.3837 0.0003 0.0877 0.1727 0.2477 0.3257 0.3887 0.0007 0.0887 0.1727 0.3307 0.3967

10 15 20 Protein analyzed (g)

25

Figure 5-2

Average absorbance values in Table 5-2 versus micrograms of protein analyzed. Averages for 0 to 20 g of protein lie on a straight line if the questionable datum 0.392 at 15 g is omitted.

Constructing a Calibration Curve

We adopt the following procedure for constructing a calibration curve: Step 1. Prepare known samples of analyte, covering a convenient range of concentrations, and measure the response of the analytical procedure to these standards. This procedure generates the data in the left half of Table 5-2. Step 2. Subtract the average absorbance (0.0993) of the blank samples from each measured absorbance to obtain corrected absorbance. The blank measures the response of the procedure when no protein is present. Step 3. Make a graph of corrected absorbance versus quantity of protein analyzed (Figure 5-3). Use the least-squares procedure to find the best straight line through the linear portion of the data, up to and including 20.0 g of protein (14 points, including the 3 corrected blanks, in the shaded portion of Table 5-2).

Absorbance of the blank can arise from the color of starting reagents, reactions of impurities, and reactions of interfering species. Blank values can vary from one set of reagents to another, but corrected absorbance should not.

86

5 Calibration Methods

0.40 0.35 Unknown beyond linear region

Calibration curve for protein analysis in Table 5-2. The equation of the solid straight line fitting the 14 data points (open circles) from 0 to 20 g, derived by the method of least squares, is y 0.016 30 ( 0.000 22) x 0.0047 ( 0.0026). The standard deviation of y is sy 0.0059. The equation of the dashed quadratic curve that fits all 17 data points from 0 to 25 g, determined by a nonlinear least-squares procedure (such as those in reference 1), is y 1.17 ( 0.21) 10 4 x2 0.018 58 ( 0.000 46) x 0.000 7 ( 0.001 0), with sy 0.0046.

Figure 5-3

0.30 Corrected absorbance 0.25 0.20 0.15 0.10 0.05 0.00 0.05 0

Unknown in linear region Quadratic calibration curve

Linear calibration curve

10 15 Amount of protein (g)

20

25

Find the slope and intercept and uncertainties with Equations 5-3, 5-4, 5-7, 5-8, and 5-9. The results are m b 0.016 30 0.0047 sm sb 0.000 22 0.0026 sy 0.0059

The equation of the linear calibration line is

Equation of calibration line: y ( sy) [m ( sm)]x [b ( sb)]

Absorbance 14243
y

( g of protein) 1442443
x

(0.016 30 )( g of protein)

0.0047

(5-12)

where y is the corrected absorbance ( observed absorbance blank absorbance). Step 4. If you analyze an unknown solution at a future time, run a blank at the same time. Subtract the new blank absorbance from the unknown absorbance to obtain the corrected absorbance.

Example Using a Linear Calibration Curve

An unknown protein sample gave an absorbance of 0.406, and a blank had an absorbance of 0.104. How many micrograms of protein are in the unknown?

SOLUTION The corrected absorbance is 0.406

g of protein absorbance 0.0047 0.016 30

0.104 0.302, which lies on the linear portion of the calibration curve in Figure 5-3. Equation 5-12 therefore becomes

Dynamic range Linear range Response

0.302 0.0047 0.016 30

18.24 g

(5-13)

c1 Analyte concentration

c2

We prefer calibration procedures with a linear response, in which the corrected analytical signal ( signal from sample signal from blank) is proportional to the quantity of analyte. Although we try to work in the linear range, you can obtain valid results beyond the linear region ( 20 g) in Figure 5-3. The dashed curve that goes up to 25 g of protein comes from a least-squares fit of the data to the equation y ax2 bx c (Box 5-1). The linear range of an analytical method is the analyte concentration range over which response is proportional to concentration. A related quantity, defined in the figure in the margin, is dynamic rangethe concentration range over which there is a measurable response to analyte, even if the response is not linear.

Before using your calculator or computer to find the least-squares straight line automatically, look at a graph of your data. This reality check gives you an opportunity to reject bad data or to decide that a straight line is not an appropriate function. It is not reliable to extrapolate any calibration curve, linear or nonlinear, beyond the measured range of standards. Measure standards in the entire concentration range of interest.

5-2 Calibration Curves

Examine your data for sensibility.

87

Propagation of Uncertainty with a Calibration Curve

In the previous example, an unknown with a corrected absorbance of y 0.302 had a protein content of x 18.24 g. What is the uncertainty in the number 18.24? A full treatment of the propagation of uncertainty gives the following result:4

Uncertainty in x ( sx )

1 Bk m sy

x2n D

a (xi ) D
2

2x a xi D

(5-14)

g of protein in unknown (18.24)

where m is the absolute value of the slope, D is from Equation 5-5, n is the number of data points (14 in Table 5-2), and k is the number of replicate measurements of the unknown. For a single measurement of the unknown, k 1 and Equation 5-14 gives sx 0.39 g. If you measure four replicate unknowns (k 4) and the average corrected absorbance is 0.302, the uncertainty is reduced from 0.39 to 0.23 g. Equation 5-14 is painful to use by hand but is readily incorporated into the spreadsheet in Figure 5-9 at the end of this chapter. Figure 5-4 shows uncertainties computed with Equation 5-14. Uncertainty is smallest near the center of the curve. Extrapolating beyond measured calibration points increases uncertainty and risks entering a region where the calibration curve is no longer linear.

xi g of protein in standards in Table 5-2 (0, 0, 0, 5.0, 5.0, 5.0, 10.0, 10.0, 10.0, 15.0, 15.0, 20.0, 20.0, 20.0) Equation C-2 in Appendix C provides a recipe for deriving Equation 5-14.

Least-squares line 6

Minimum uncertainty in x is at the centroid 4 y Data point Centroid of measured data (x, y )

Uncertainty in x increases as distance from centroid increases 0 4 2 0 2 x 4 6 8 10

Uncertainty in x computed with Equation 5-14 is smallest near the centroid (x , y ) of the least-squares calibration line. The four dark circles are the data points from which the least-squares line was calculated [(1,2.5), (3,2.8), (4,4.5), (6,5)]. The centroid is at the average value of x and y (3.5, 3.7). At the upper and lower extremes in the figure, the uncertainty in x is 60% greater than the uncertainty in the middle.

Figure 5-4

Box 5-1

Using a Nonlinear Calibration Curve

Substituting a 1.17 10 4, b 0.375 7 into these equations gives 0.018 58, and c

Consider an unknown whose corrected absorbance of 0.375 lies beyond the linear region in Figure 5-3. We can fit all the data points in Figure 5-3 with the quadratic equation

135 g

23.8 g

1.17

10

x2

0.018 58 x

0.000 7 (A)

whose coefficients were found by the method in Supplementary Problem S17-33 at www.whfreeman.com/qca. To find the quantity of protein, substitute the measured absorbance into Equation A:

0.375 1.17

1.17 10
4

10 x2 ax2

x2

0.018 58 x 0.375 7

0.000 7 0

Figure 5-3 tells us that the correct choice is 23.8 g, not 135 g. Estimating uncertainty. How can we estimate the uncertainty in the quantity of protein if the absorbance of the unknown is 0.375 0.006? A simple way to do this is to substitute the upper and lower limits (0.381 and 0.369) back into Equation A:

This can be rearranged to

Upper limit: 0.381 0.018 58 x Lower limit: 0.369

2b2

1.17 1.17

which is a quadratic equation of the form

2b2

bx 4ac

0 b

10 4 x2 0.018 58 x 0.000 7 X x 24.2 g 10 4 x2 0.018 58 x 0.000 7 X x 23.3 g

whose two possible solutions are

2a

4ac

A reasonable way to express the answer is 23.8 0.5 g. (We have neglected the effect of uncertainty in the parameters a, b, and c.)

2a

8 Relative mass spectral peak area Matrix = reagent water 6

5-3 Standard Addition5

In standard addition, known quantities of analyte are added to the unknown. From the increase in signal, we deduce how much analyte was in the original unknown. This method requires a linear response to analyte. Standard addition is especially appropriate when the sample composition is unknown or complex and affects the analytical signal. The matrix is everything in the unknown, other than analyte. We define a matrix effect as a change in the analytical signal caused by anything in the sample other than analyte. Figure 5-5 shows a strong matrix effect in the analysis of perchlorate (ClO4 ) by mass spectrometry (described in Chapter 22). Perchlorate at a level above 18 g/L in drinking water is of concern because it can reduce thyroid hormone production. Standard solutions of ClO4 in pure (reagent) water gave the upper calibration curve in Figure 5-5. The response to standard solutions with the same concentrations of ClO4 in groundwater was 15 times less, as shown in the lower curve. Reduction of the ClO4 signal is a matrix effect attributed to other anions present in the groundwater. Because different groundwaters have different concentrations of many anions, there is no way to construct a calibration curve for this analysis that would apply to more than one specific groundwater. Hence, the method of standard addition is required. When we add a small volume of concentrated standard to an existing unknown, we do not change the concentration of the matrix very much. The assumption in standard addition is that the matrix has the same effect on added analyte as it has on the original analyte in the unknown. Consider a standard addition in which a sample with unknown initial concentration of analyte [X]i gives a signal intensity IX. Then a known concentration of standard, S, is added to an aliquot of the sample and a signal IS X is observed for this second solution. Addition of standard to the unknown changes the concentration of the original analyte because of dilution. Lets call the diluted concentration of analyte [X]f, where f stands for final. We designate the concentration of standard in the final solution as [S]f. (Bear in mind that the chemical species X and S are the same.)

2 Matrix = groundwater 0 0 20 40 60 80

Perchlorate (g/L)

Figure 5-5

Calibration curves for perchlorate in pure water and in groundwater. [Data from C. J. Koester, H. R.

Beller, and R. U. Halden, Analysis of Perchlorate in Groundwater by Electrospray Ionization Mass Spectrometry/Mass Spectrometry, Environ. Sci. Technol. 2000, 34, 1862.]

The matrix affects the magnitude of the analytical signal. In standard addition, all samples are in the same matrix.

88

Signal is directly proportional to analyte concentration, so

89
5-3 Standard Addition

Concentration of analyte in initial solution Concentration of analyte plus standard in final solution Xi Sf Xf IX IS
X

signal from initial solution signal from final solution

Standard addition equation:

(5-15)

Derivation of Equation 5-15: IX k [X]i, where k is a constant of proportionality IS X k([S]f [X]f), where k is the same constant Dividing one equation by the other gives IX IS
X

For an initial volume V0 of unknown and added volume Vs of standard with concentration [S]i, the total volume is V V0 Vs and the concentrations in Equation 5-15 are

X ia

V0 b V

S ia

VS b V

(5-16)

The quotient (initial volume/final volume), which relates final concentration to initial concentration, is called the dilution factor. It comes directly from Equation 1-3.

k( 3 S4 f

k 3X4 i

3 X 4 f)

3S 4 f

3X4 i

3X 4 f

By expressing the diluted concentration of analyte, [X]f, in terms of the initial concentration of analyte, [X]i, we can solve for [X]i, because everything else in Equation 5-15 is known.

Example Standard Addition

Serum containing Na gave a signal of 4.27 mV in an atomic emission analysis. Then 5.00 mL of 2.08 M NaCl were added to 95.0 mL of serum. This spiked serum gave a signal of 7.98 mV. Find the original concentration of Na in the serum.
A spike is analyte deliberately added to a sample.

SOLUTION From Equation 5-16, the final concentration of Na after dilution with the
standard is [X]f [X]i(V0 /V) [X]i(95.0 mL/100.0 mL). The final concentration of added standard is [S]f [S]i(Vs/V) (2.08 M)(5.00 mL/100.0 mL) 0.104 M. Equation 5-15 becomes

Na 0.104 M

0.950 Na

4.27 mV 7.98 mV

Na

0.113 M

A Graphic Procedure for Standard Addition with Constant Total Volume

An excellent way to carry out standard addition is illustrated in Figure 5-6. Pipet equal volumes of unknown into several volumetric flasks. Then add increasing volumes of standard to each flask and dilute to the mark. Every flask now contains the same concentration of unknown and differing concentrations of standard. (Remember that the standard is the same substance as the unknown.) The next step is to analyze the solutions and construct the graph in Figure 5-7. The x-axis is the concentration of added standard after it has been mixed with sample. The x-intercept of the extrapolated line is the concentration of unknown ([X]f) after it has been diluted to the final volume. In Figure 5-7, this concentration is 0.042 0 M. If the original unknown had been diluted by a factor of 10.0 (from 5.00 up to 50.0 mL) by standard additions, then the original unknown was 0.420 M. The most useful range of standard additions should increase the analytical signal to between 1.5 and 3 times its original value. If you fit the five points in Figure 5-7 with a least-squares line, the uncertainty in the x-intercept (the standard deviation of [X]f) is4,6

The equation of the line in Figure 5-7 is y mx b. The x-intercept is obtained by setting y 0: 0 x mx b b/m

Standard deviation of intercept

m 2D sy

2 n(intercept) 2

2(intercept) xi

(x2 ) i

(5-17)

Uncertainty for standard addition

where sy is the standard deviation of y (Equation 5-7), m is the slope of the least-squares line (Equation 5-3), D is given by Equation 5-5, n is the number of data points (5), intercept is the x-intercept, and xi are the values of x for the 5 points in Figure 5-7.

90
5 Calibration Methods
1

Place 5 mL of unknown in each flask

Add 0, 5, 10, 15, or 20 mL of standard

Figure 5-6 Standard addition experiment. Suppose that the stock solution that is added to each flask contains 0.200 M standard. At the end of the procedure, flask 1 contains no added standard. Flask 2 contains 0.020 0 M standard (i.e., 5.00 mL diluted to 50.0 mL). Flasks 3, 4, and 5 contain 0.040 0, 0.060 0, and 0.080 0 M standard, respectively.

Fill each flask to the 50-mL mark and mix

Analytical signal

ta ob rd gs nda in a ad st Re ded ad

d ine

wi

th

Figure 5-7

Graphical treatment of standard addition experiment in Figure 5-6. Standard additions should increase the analytical signal to between 1.5 and 3 times its original value (i.e., B 0.5A to 2A).

Concentration of unknown, [X]f

Reading of unknown without added standard

0.04

0.02

0.02 0.04 0.06 0.08 Concentration of added analyte, [S]f (M)

A Graphic Procedure for Standard Addition with Varying Volume

In Figure 5-6, we needed a separate flask for each standard addition because the chemical analysis consumes solution. In atomic spectroscopy or mass spectrometry, some solution from each flask will be sucked into the spectrometer and used up. In some other analyses, we measure a property of the analyte without consuming solution. For example, we could place electrodes in a known volume of sample and make an electrical measurement. Then we make a standard addition, which increases the total volume, and measure the electrode response again. Then we repeat the process

several times until the original signal has increased by a factor of 1.5 to 3. The volume of solution increases in each successive measurement. To treat this case of increasing volume, we plug [X]f and [S]f from Equations 5-16 into Equation 5-15 and rearrange to find

91
5-4 Internal Standards
Xa

For successive standard additions to one solution:

V b V0 1 2 3 4 4 IS
Function to plot on y-axis

Xa

IX

VS IX 3S4 i a b 3X4 i V0 123

Function to plot on x-axis

(5-18)

Plot IS

VS V b versus [S]i a b V0 V0

x-intercept original concentration of unknown Uncertainty of intercept is given by Equation 5-17

A graph of IS X(V/V0) (the corrected response) on the y-axis versus [S]i(VS/V0) on the x-axis is a straight line. The right side of Equation 5-18 is 0 when [S]i(VS/V0) [X]i. The magnitude of the intercept on the x-axis is the original concentration of unknown, [X]i, whose uncertainty is given by Equation 5-17.

5-4 Internal Standards

An internal standard is a known amount of a compound, different from analyte, that is added to the unknown. Signal from analyte is compared with signal from the internal standard to find out how much analyte is present. Internal standards are especially useful for analyses in which the quantity of sample analyzed or the instrument response varies slightly from run to run for reasons that are difficult to control. For example, gas or liquid flow rates that vary by a few percent in a chromatography experiment (Figure 0-4) could change the detector response. A calibration curve is only accurate for the one set of conditions under which it was obtained. However, the relative response of the detector to the analyte and standard is usually constant over a wide range of conditions. If signal from the standard increases by 8.4% because of a change in flow rate, signal from the analyte usually increases by 8.4% also. As long as the concentration of standard is known, the correct concentration of analyte can be derived. Internal standards are widely used in chromatography because the small quantity of sample solution injected into the chromatograph is not reproducible. Internal standards are also desirable when sample loss can occur during sample preparation steps prior to analysis. If a known quantity of standard is added to the unknown prior to any manipulations, the ratio of standard to analyte remains constant because the same fraction of each is lost in any operation. To use an internal standard, we prepare a known mixture of standard and analyte to measure the relative response of the detector to the two species. In the chromatogram in Figure 5-8, the area under each peak is proportional to the concentration of the species injected into the column. However, the detector generally has a different response to each component. For example, if both the analyte (X) and internal standard (S) have concentrations of 10.0 mM, the area under the analyte peak might be 2.30 times greater than the area under the standard peak. We say that the response factor, F, is 2.30 times greater for X than for S.
In standard addition, the standard is the same substance as the analyte. An internal standard is a different substance from the analyte. When the relative response of an instrument to analyte and standard remains constant over a range of concentrations, we say there is a linear response. For critical work, this assumption should be verified because it is not always true.7

Detector signal

5 Time (min)

10

Response factor:

Area of analyte signal Concentration of analyte AX 3X4

Fa Fa

area of standard signal b concentration of standard

(5-19)

Figure 5-8

AS b 3S4

[X] and [S] are the concentrations of analyte and standard after they have been mixed together. Equation 5-19 is predicated on linear response to both the analyte and the standard.

Chromatographic separation of unknown (X) and internal standard (S). A known amount of S was added to the unknown. The relative areas of the signals from X and S allow us to find out how much X is in the mixture. It is necessary first to measure the relative response of the detector to each compound. This figure shows a chromatogram, which is a graph of detector response versus time.

92
5 Calibration Methods
If the detector responds equally to standard and analyte, F 1. If the detector responds twice as much to analyte as to standard, F 2. If the detector responds half as much to analyte as to standard, F 0.5.

Example Using an Internal Standard

In a preliminary experiment, a solution containing 0.083 7 M X and 0.066 6 M S gave peak areas of AX 423 and AS 347. (Areas are measured in arbitrary units by the instruments computer.) To analyze the unknown, 10.0 mL of 0.146 M S were added to 10.0 mL of unknown, and the mixture was diluted to 25.0 mL in a volumetric flask. This mixture gave the chromatogram in Figure 5-8, for which AX 553 and AS 582. Find the concentration of X in the unknown.

SOLUTION First use the standard mixture to find the response factor in Equation 5-19:
Standard mixture: 423 0.083 7 3S 4 AS AX Fa b 3X4 3S4 347 Fa b 1 0.066 6 10.0 b (0.146 M) a 25.0 1 2 3 123 4 4
Initial Dilution concentration factor

0.9700

In the mixture of unknown plus standard, the concentration of S is

initial volume The dilution factor final volume converts initial concentration to final concentration.

0.058 4 M

Using the known response factor, we substitute back into Equation 5-19 to find the concentration of unknown in the mixture:

Unknown mixture: 553 3X4

AS AX Fa b 3X4 3S4 582 0.9700 a b 1 3X4 0.058 4

0.057 21 M

Because X was diluted from 10.0 to 25.0 mL when the mixture with S was prepared, the original concentration of X in the unknown was (25.0 mL/10.0 mL)(0.057 21 M) 0.143 M.

5-5
A living version of the spreadsheet and graph in Figure 5-9 can be found at the Web site for this book: www.whfreeman.com/qca.

A Spreadsheet for Least Squares

How to use TRENDLINE to add the straight line to the least-squares graph.

Figure 5-9 translates the least-squares computations of Table 5-1 into a spreadsheet. Enter values of x and y in columns B and C, and the total number of points (n 4) in cell A5. Compute the products xy and x2 in columns D and E. Compute the sums of columns B through G in row 9. For example, you can find the sum of xy values in cells D4 through D7 with the statement SUM(D4:D7). The least-squares parameters D, m, and b are computed in cells A12, A14, and A16 by using Equations 5-3 through 5-5. The vertical deviations, d, in column F make use of the slope and intercept. Column G contains the squares of the deviations. Compute the standard deviations sy sm, and sb in cells B12, B14, and B16, using Equations 5-7 through 5-9. Document all formulas in the spreadsheet. The bottom of the spreadsheet uses Equation 5-14 to evaluate uncertainty in a derived value of x. Enter a measured value of y in cell A22 and compute the derived values of x and its uncertainty in cells A24 and G24. This spreadsheet tells us that for an observed value of y 2.72 in Figure 5-1, the value of x is 2.23 0.37. If replicate values of y are measured, then cell G22 should contain the number of measurements and cell A22 should have the average value of y. The graph in Figure 5-9 is important because it tells us quickly how well the data follow a straight line. Section 2-11 tells how to create a graph showing data points. To add a straight line, click on one data point and they will all be highlighted. Then go to the INSERT menu and select TRENDLINE. In the window that appears, select Linear. Go to the CHART menu and select ADD TRENDLINE. In some versions of Excel there is no CHART menu. In this case, go to Options in the TRENDLINE box and select Display Equation on

Summary
A 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 b= 1.34615 B C D E F G H I J K L

93

Least-Squares Spreadsheet Number of points (n) 4 x 1 3 4 6 14 D= 52 m= 0.61538 std dev(y)= 0.19612 std dev(m)= 0.05439 std dev(b)= 0.21414 y 2 3 4 5 14 xy 2 9 16 x^2 1 9 16 d 0.03846 0.19231 0.19231 0.03846 0.0000 d^2 0.00148 0.03698 0.03698 0.00148 7.7E-02
6 y = 0.6154 + 1.3462 5 Least-Squares Straight Line

36 30 Column Sums 57 62

A12 = $A$5*E9-B9*B9 A14 = (D19*$A$5-B9*C9)/$A$12 A16 = (E9*C9-D9*B9)/$A$12 B12 = SQRT(G9/($A$5-2)) B14 = $B$12*SQRT($A$5/$A$12) B16 = $B$12*SQRT(E9/$A$12) F4 = C4-$A$14*B4-$A$16

Finding uncertainty in x with Equation 5-14:

0

Measured y = 2.72 Derived x = 2.23250

Number of replicate values of y measured (k) = 1 Uncertainty in x = 0.3735

2 x

Derived x in cell A24 = (A22-A16)/A14 Uncertainty in x cell G24 = ($B$12/ABS($A$14))*SQRT((1/$G$22) +$A$24^2*$A$5/$A$12+$E$9/$A$12 2*$A$24*$B$9/$A$12) Spreadsheet for linear least-squares analysis.

Figure 5-9

Chart. When you click OK, the least-squares straight line and its equation appear on the graph. Double click on the line and you can adjust its thickness and appearance. Double clicking on the equation allows you to modify its format. If you want to extend the straight line, double click on it and select Options. In the Forecast box, you can extend the trendline Forward and Backward as far as you like.

Terms to Understand
blank solution calibration curve determinant dilution factor dynamic range intercept internal standard linear range linear response matrix matrix effect method of least squares response factor slope standard addition standard solution

Summary
A calibration curve shows the response of a chemical analysis to known quantities (standard solutions) of analyte. When there is a linear response, the corrected analytical signal ( signal from sample signal from blank) is proportional to the quantity of analyte. Blank solutions are prepared from the same reagents and solvents used to prepare standards and unknowns, but blanks have no intentionally added analyte. The blank tells us the response of the procedure to impurities or interfering species in the reagents. The blank value is subtracted from measured values of standards prior to constructing the calibration curve. The blank value is subtracted from the response of an unknown prior to computing the quantity of analyte in the unknown.

94

5 Calibration Methods
used to construct the graph in Figure 5-7. The x-intercept in Figure 5-7 is the concentration of analyte in the sample to which no internal standard was added. For multiple standard additions to one solution with increasing volume, we graph Equation 5-18 and use the intercept to find the concentration of unknown. An internal standard is a known amount of a compound, different from analyte, that is added to the unknown. Signal from analyte is compared with signal from the internal standard to find out how much analyte is present. Internal standards are especially useful when the quantity of sample analyzed is not reproducible, when instrument response varies from run to run, or when sample losses occur during sample preparation. The response factor in Equation 5-19 is the relative response to analyte and standard.

The method of least squares is used to determine the equation of the best straight line through experimental data points. Equations 5-3 to 5-5 and 5-7 to 5-9 provide the leastsquares slope and intercept and their standard deviations. Equation 5-14 estimates the uncertainty in a quantity measured from a calibration curve. A spreadsheet greatly simplifies least-squares calculations. A standard addition is a known quantity of analyte added to an unknown to increase the concentration of analyte by a known amount. Standard additions are especially useful when matrix effects are important. A matrix effect is a change in the analytical signal caused by anything in the sample other than analyte. You should be able to use Equation 5-15 to compute the quantity of analyte in a standard addition experiment. Multiple standard additions with a constant total volume are

Exercises
5-A. Calibration curve. (You can do this problem completely with your calculator, but it is much more easily done by the spreadsheet in Figure 5-9.) In the Bradford protein determination, the color of a dye changes from brown to blue when it binds to protein. The intensity of blue color (measured by absorbance of light at a wavelength of 595 nm) is proportional to protein concentration. Protein ( g): 0.00 9.36 18.72 Absorbance at 595 nm: 0.466 0.676 0.883 28.08 1.086 37.44 1.280 5-B. Standard addition. An unknown sample of Ni2 gave a current of 2.36 A in an electrochemical analysis. When 0.500 mL of solution containing 0.028 7 M Ni2 was added to 25.0 mL of unknown, the current increased to 3.79 A. (a) Denoting the initial, unknown concentration as [Ni2 ]i, write an expression for the final concentration, [Ni2 ]f, after 25.0 mL of unknown were mixed with 0.500 mL of standard. Use the dilution factor for this calculation. (b) In a similar manner, write the final concentration of added standard Ni2 , designated as [S]f. (c) Find [Ni2 ]i in the unknown. 5-C. Internal standard. A solution was prepared by mixing 5.00 mL of unknown (element X) with 2.00 mL of solution containing 4.13 g of standard (element S) per milliliter and diluting to 10.0 mL. The measured signal ratio in an atomic absorption experiment was (signal due to X) / (signal due to S) 0.808. In a separate experiment, it was found that for equal concentrations of X and S, the signal due to X was 1.31 times more intense than the signal due to S. Find the concentration of X in the unknown.

(a) Determine the equation of the least-squares straight line through these points in the form y [m( sm)]x [b( sb)] with a reasonable number of significant figures. (b) Make a graph showing the experimental data and the calculated straight line. (c) An unknown protein sample gave an absorbance of 0.973. Calculate the number of micrograms of protein in the unknown and estimate its uncertainty.

Problems
Linear Least Squares
5-1. A straight line is drawn through the points (3.0, 3.87 104), (10.0, 12.99 104), (20.0, 25.93 104), (30.0, 38.89 104), and (40.0, 51.96 104), using the method of least squares. The results are m 1.298 72 104, b 256.695, sm 13.190, sb 323.57, and sy 392.9. Express the slope and intercept and their uncertainties with reasonable significant figures. 5-2. Here is a least-squares problem that you can do by hand with a calculator. Find the slope and intercept and their standard deviations for the straight line drawn through the points (x,y) (0,1), (2,2), and (3,3). Make a graph showing the three points and the line. Place error bars ( sy) on the points. 5-3. Set up a spreadsheet like the one in Figure 5-9 to reproduce results in Figure 5-9. Add error bars: Double click on a data point on the graph and select Y Error Bars. Check Custom and enter the value of sy in each box for the and error. Better yet, enter the cell containing sy in both boxes. 5-4. Excel LINEST function. Enter the data from Problem 5-1 in a spreadsheet and use the LINEST function to find

Problems
the slope and intercept and standard deviations. Use Excel to draw a graph of the data and insert a TRENDLINE.

95

(c) Estimate the uncertainty in the quantity of protein by using Equation 5-14 in your spreadsheet. 5-10. Here are mass spectrometric signals for known concentrations of methane in H2: CH4 (vol %): 0 0.062 0.122 0.245 0.486 0.971 1.921 Signal (mV): 9.1 47.5 95.6 193.8 387.5 812.5 1 671.9 (a) Subtract the blank value (9.1) from all other values. Then use the method of least squares to find the slope and intercept and their uncertainties. Construct a calibration curve. (b) Replicate measurements of an unknown gave 152.1, 154.9, 153.9, and 155.1 mV, and a blank gave 8.2, 9.4, 10.6, and 7.8 mV. Subtract the average blank from the average unknown to find the average corrected signal for the unknown. (c) Find the concentration of the unknown and its uncertainty. 5-11. The figure below gives replicate measurements of As(III) concentration by an electrochemical method. (a) Using a millimeter ruler, measure each peak height to the nearest 0.1 mm. Noting the length that corresponds to 200 nA in the figure, make a table showing the observed current (nA) for each concentration ( M) of As(III). The blanks appear to be near 0, so we will disregard them in this problem. (b) Construct a calibration curve with 24 points (AD) and find the slope and intercept and their uncertainties, using the method of least squares. (c) Calculate the concentration (and its uncertainty) of As(III) in an unknown that gave a mean current of 501 nA from six measurements.

Calibration Curves
5-5. Explain the statement: The validity of a chemical analysis ultimately depends on measuring the response of the analytical procedure to known standards. 5-6. Suppose that you carry out an analytical procedure to generate a calibration curve like that shown in Figure 5-3. Then you analyze an unknown and find an absorbance that gives a negative concentration for the analyte. What does this mean? 5-7. Using the calibration curve in Figure 5-3, find the quantity of unknown protein that gives a measured absorbance of 0.264 when a blank has an absorbance of 0.095. 5-8. Consider the least-squares problem in Figure 5-1. (a) Suppose that a single new measurement produces a y value of 2.58. Find the corresponding x value and its uncertainty. (b) Suppose you measure y four times and the average is 2.58. Calculate the uncertainty based on four measurements, not one. 5-9. Consider the linear calibration curve in Figure 5-3, which is derived from the 14 corrected absorbances in the shaded region at the right side of Table 5-2. (a) Create a least-squares spreadsheet to compute the equation of the line and the standard deviations of the parameters (sy, sm, sb). (b) Suppose that you find absorbance values of 0.265, 0.269, 0.272, and 0.258 for four identical samples of unknown and absorbances of 0.099, 0.091, 0.101, and 0.097 for four blanks. Find the corrected absorbance by subtracting the average blank from the average absorbance of the unknown. Calculate the amount of protein in the unknown.

200 nA

B Current

Figure for Problem 5-11: Electrochemical analysis of As(III). Replicate samples correspond to (A) 20 M, (B) 30 M, (C) 40 M, (D) 50 M As(III), and (E) blanks. [From I. G. R. Gutz,
O. L. Angnes, and J. J. Pedrotti, Adaptation of Poly(tetrafluoroethylene) Tips to Mercury Drop Electrodes and Evaluation by Flow Injection Analysis, Anal. Chem. 1993, 65, 500.] E Time

96

5 Calibration Methods
5-15. Confidence interval for calibration curve. To use a calibration curve based on n points, we measure a new value of y and calculate the corresponding value of x. The one-standarddeviation uncertainty in x, sx is given by Equation 5-14. We can express a confidence interval for x, using Students t:

5-12. Nonlinear calibration curve. Following the procedure in Box 5-1, find how many micrograms ( g) of protein are contained in a sample with a corrected absorbance of 0.350 in Figure 5-3. 5-13. Logarithmic calibration curve. Calibration data spanning five orders of magnitude for an electrochemical determination of p-nitrophenol are given in the table. (The blank has already been subtracted from the measured current.) If you try to plot these data on a linear graph extending from 0 to 310 g/mL and from 0 to 5 260 nA, most of the points will be bunched up near the origin. To handle data with such a large range, a logarithmic plot is helpful. p-Nitrophenol ( g/mL) 0.010 0 0.029 9 0.117 0.311 1.02 Current (nA) 0.215 0.846 2.65 7.41 20.8 p-Nitrophenol ( g/mL) 3.00 10.4 31.2 107 310 Current (nA) 66.7 224 621 2 020 5 260

Confidence interval

tsx

where t is taken from Table 4-2 for n 2 degrees of freedom. A calibration curve based on n 10 known points was used to measure the protein in an unknown. Based on Equation 5-14, the results were protein 15.22 ( 0.46) g. Find the 90% and 99% confidence intervals for protein in the unknown.

Standard Addition
5-16. Why is it desirable in the method of standard addition to add a small volume of concentrated standard rather than a large volume of dilute standard? 5-17. An unknown sample of Cu2 gave an absorbance of 0.262 in an atomic absorption analysis. Then 1.00 mL of solution containing 100.0 ppm ( g/mL) Cu2 was mixed with 95.0 mL of unknown, and the mixture was diluted to 100.0 mL in a volumetric flask. The absorbance of the new solution was 0.500. (a) Denoting the initial, unknown concentration as [Cu2 ]i, write an expression for the final concentration, [Cu2 ]f, after dilution. Units of concentration are ppm. (b) In a similar manner, write the final concentration of added standard Cu2 , designated as [S]f. (c) Find [Cu2 ]i in the unknown. 5-18. Europium is a lanthanide element found in parts per billion levels in natural waters. It can be measured from the intensity of orange light emitted when a solution is illuminated with ultraviolet radiation. Certain organic compounds that bind Eu(III) are required to enhance the emission. The figure below shows standard addition experiments in which
Emission peak area

Data from Figure 4 of L. R. Taylor, Am. Lab., February 1993, p. 44.

(a) Make a graph of log(current) versus log(concentration). Over what range is the log-log calibration linear? (b) Find the equation of the line in the form log(current) m log(concentration) b. (c) Find the concentration of p-nitrophenol corresponding to a signal of 99.9 nA. 5-14. Use a spreadsheet to prepare a graph analogous to Figure 5-4 from the four data points in Table 5-1. Use Equation 5-14 to compute the uncertainty in x for y 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 5.5, and 6. Plot these uncertainties as error bars in the x direction as shown in Figure 5-4. How much greater is the uncertainty in x for y 1 than for the centroid (y 3.5)?

40

30

Pond water

20

14.6 mL

10 6.0 mL

Tap water

Figure for Problem 5-18: Standard addition of Eu(III) to pond water or tap water. [Data from A. L. Jenkins
and G. M. Murray, Enhanced Luminescence of Lanthanides, J. Chem. Ed. 1998, 75, 227.] 15 10

10

15

Added volume of standard (mL)

Problems
10.00 mL of sample and 20.00 mL containing a large excess of organic additive were placed in 50-mL volumetric flasks. Then Eu(III) standards (0, 5.00, 10.00, or 15.00 mL) were added and the flasks were diluted to 50.0 mL with H2O. Standards added to tap water contained 0.152 ng/mL (ppb) of Eu(III), but those added to pond water were 100 times more concentrated (15.2 ng/mL). (a) From the x-intercepts in the graph, calculate the concentration of Eu(III) (ng/mL) in pond water and tap water. (b) In the case of tap water, the emission peak area increases by 4.61 units when 10.00 mL of 0.152 ng/mL standard are added. This response is 4.61 units / 1.52 ng 3.03 units per ng of Eu(III). For pond water, the response is 12.5 units when 10.00 mL of 15.2 ng/mL standard are added, or 0.082 2 units per ng. How would you explain these observations? Why was standard addition necessary for this analysis? 5-19. Standard addition graph for constant total volume. Students performed an experiment like that in Figure 5-6 in which each flask contained 25.00 mL of serum, varying additions of 2.640 M NaCl standard, and a total volume of 50.00 mL. Flask 1 2 3 4 5 Volume of standard (mL) 0 1.000 2.000 3.000 4.000 Na atomic emission signal (mV) 3.13 5.40 7.89 10.30 12.48

97

Internal Standards
5-21. State when standard additions and internal standards, instead of a calibration curve, are desirable, and why. 5-22. A solution containing 3.47 mM X (analyte) and 1.72 mM S (standard) gave peak areas of 3 473 and 10 222, respectively, in a chromatographic analysis. Then 1.00 mL of 8.47 mM S was added to 5.00 mL of unknown X, and the mixture was diluted to 10.0 mL. This solution gave peak areas of 5 428 and 4 431 for X and S, respectively. (a) Calculate the response factor for the analyte. (b) Find the concentration of S (mM) in the 10.0 mL of mixed solution. (c) Find the concentration of X (mM) in the 10.0 mL of mixed solution. (d) Find the concentration of X in the original unknown. 5-23. Chloroform is an internal standard in the determination of the pesticide DDT in a polarographic analysis in which each compound is reduced at an electrode surface. A mixture containing 0.500 mM chloroform and 0.800 mM DDT gave signals of 15.3 A for chloroform and 10.1 A for DDT. An unknown solution (10.0 mL) containing DDT was placed in a 100-mL volumetric flask and 10.2 L of chloroform (FM 119.39, density 1.484 g/mL) were added. After diluting to the mark with solvent, polarographic signals of 29.4 and 8.7 A were observed for the chloroform and DDT, respectively. Find the concentration of DDT in the unknown. 5-24. Verifying constant response for an internal standard. When we develop an analytical method using an internal standard, it is important to verify that the response factor is constant. Data are shown below for a chromatographic analysis of naphthalene (C10H8) using deuterated naphthalene (C10D8 in which D is the isotope 2H) as an internal standard. The two compounds emerge from the column at almost identical times and are measured by a mass spectrometer, which distinguishes them by molecular mass. From the definition of response factor in Equation 5-19, we can write

(a) Prepare a standard addition graph and find the concentration of Na in the serum. (b) Use Equation 5-17 to find the uncertainty in the answer to part (a). 5-20. Standard addition graph for variable volume. An assay for substance X is based on its ability to catalyze a reaction that produces radioactive Y. The quantity of Y produced in a fixed time is proportional to the concentration of X in the solution. An unknown containing X in a complex, unknown matrix with an initial volume of 50.0 mL was treated with increments of standard 0.531 M X and the following results were obtained: Volume of added X ( L): 0 100.0 200.0 300.0 3 266 400.0 4 010

Area of analyte signal Area of standard signal

Fa

concentration of analyte b concentration of standard

Prepare a graph of peak area ratio (C10H8/C10D8) versus concentration ratio (C10H8/C10D8) and find the slope, which is the response factor. Evaluate F for each of the 3 samples and find the standard deviation of F to see how constant it is. Sample 1 2 3 C10H8 (ppm) 1.0 5.0 10.0 C10D8 (ppm) 10.0 10.0 10.0 C10H8 peak area 303 3 519 3 023 C10D8 peak area 2 992 6 141 2 819

Counts/min of radioactive Y: 1 084 1 844 2 473

(a) Prepare a standard addition graph based on Equation 5-18 and find the concentration of X in the original unknown. (b) Use Equation 5-17 to find the uncertainty in the answer to (a).

The volume of solution injected into the column was different in all three runs.

98

5 Calibration Methods
Figure 5-7 is 0.0423 0.0021 M (where 0.0021 is the standard deviation computed with Equation 5-17). Find the 90% and 99% confidence intervals expressed in the form [X]f 0.0423 ( ?). As in any linear least-squares procedure, the number of degrees of freedom is n 2.

5-25. Confidence interval for standard addition. For a standard addition graph (Figure 5-7), Equation 5-17 gives the onestandard-deviation uncertainty in the x-intercept. If there are n data points in the line (including the point with no added standard), the confidence interval for the x-intercept is t (standard deviation of x-intercept). Suppose that the x-intercept in

Experiments
S. Pandey, M. E. R. McHale, K. S. Coym, and W. E. Acree, Jr., Bilinear Regression Analysis as a Means to Reduce Matrix Effects in Simultaneous Spectrophotometric Determination of CrIII and CoII, J. Chem. Ed. 1998, 75, 878.