Exercise No.

Method Validation

Charles Michael B. Umerez

 I. Introduction

Method validation is the process of proving that an analytical method is acceptable for its
intended purpose (Harris, 2007). Method validation is important especially in analytical measurement,
quality management systems, conformity assessments and a laboratory should implement this to allow it
to produce a reliable analytical data. It is one of the measures that is universally recognized as a necessary
part of a comprehensive system of quality assurance in analytical chemistry (Thompson et al, 1999).

Full validation is needed and internationally accepted protocols have been established for the full
validation of a method analysis. Some of this international organizations are International Harmonized
Protocol and ISO. These protocols/standards require a minimum number of laboratories and test
materials to be included in the collaborative trial to validate fully the analytical method.

Method validation follows basic principles. First principle is specification and scope of validation. In
general, validation should check that the method performs adequately for the purpose throughout the
range of analyte concentrations and test materials to which it is applied. Second principle is testing
assumptions. Validation studies act as an objective test of any assumptions on which an analytical method
is based. Third principle the sources of error in analysis. Errors in analytical measurements arise from
different sources and at different levels of organization Lastly, the fourth principle is the method and
laboratory effects. It is critically important in single-laboratory method validation to take account of
method bias and the laboratory effect.

Requirements for method validation includes measuring the method specificity, linearity, accuracy,
precision, range, limit of detection, limit of quantitation and robustness.

Specificity is defined as the ability of an analytical method to distinguish the analyte from everything else
that might be in the sample. Linearity measures how well a calibration curve follows a straight line.
Accuracy is defined as the closeness of the value to the true value, or the nearness. Precision is the
reproducibility of the result, or how close are the data collected from each other. Range is defined as the
concentration interval over which linearity, accuracy, and precision are all acceptable. Limit of detection
is the smallest quantity of analyte that is significantly different from the blank. Lastly, robustness is defined
as the ability of an analytical method to be unaffected by small, deliberate changes in operating
parameters (Harris, 2007).

I. Methodology

A. Sample Preparation

The sample that was analyzed was mountain dew and 5- mL of the sample was placed in a beaker and
warmed on a hot plate to expel CO2. The particles were removed by filtration and the sample was allowed
to cool after to room temperature.

B. Selectivity

For the determination of UV spectra, 25 mL of (a) Blank, (b) 6 mg/L benzoic acid standard solution, and (c)
diluted sample (1:25) was prepared. For the sample, 2.5 mL of 0.10 M HCl was added before diluting with
distilled water The UV spectrum (200-350) nm of the solutions were obtained. An overlapped spectrum
of the blank, standard, and sample was obtained. The peak absorbance was determined and was the
spectra was checked for interference peaks of impurities in the sample

For caffeine interference, 50 mL of 0, 2, 4, 6, 8, and 10 mg/L benzoic acid standards were prepared. The
standards were added with 5 mL of 0.10 M HCl before diluting with distilled water. Another set of 25 mL
standards containing a mixture of benzoic acid and caffeine was prepared; each standard solution
containing the same amount of the two compounds: 0, 2, 4, 6, 8 and 10 mg/L. Before diluting with distilled
water, 2.5 mL of 0.10 M HCl was added. The absorbances of all the standards at 230 nm was read.
Calibration curves for the benzoic acid standards and the mixture of benzoic acid and caffeine was
constructed.

C.LOD/LOQ

The sample was diluted in a 1:100 ratio and 10 diluted samples were prepared, 100 ml each. The sample
was added with 10 mL of 0.10 M HCl before diluting with distilled water

The absorbance of each solution was determined. The standard deviation of the readings were calculated
and the LOD and LOQ were estimated.

D. Linearity/Working Range

Using the stock solution (92mg/mL), six 50-mL calibration standards were prepared with the following
concentrations: 0,1,2,4,6,8, and 10 mg/mL. The absorbance reading at 230 nm of each standard was
determined in a randomized order. Three absorbance readings were obtained for each standard.

The net readings were determined and the net absorbance readings vs. the concentration was plotted.

E. Trueness/ Recovery

Six 25 mL solutions of the sample with dilution factor of 1:100 were spiked with 1 mg/L benzoic acid (low)
and 4 mg/L benzoic acid (high) were prepared. Also, six 25 mL unspiked samples were prepared/ The
absorbance at 230 nm of each of the spiked and unspiked sample was determined.

Calibration curve from the selectivity assessment was used to calculate the concentration of each sample.
Grubb's test was used to check for outliers. The %Recover and average %recovery were determined. Using
t-test, the significance of the differences of average recoveries was obtained at two different spiking
levels.

F. Repeatability and within-laboratory precision

Six replicates of the test samples with dilution factor of 1:100 were prepared. The samples were added
with 2.5 mL of 0.10 M HCl before it was diluted with water. The absorbance reading at 230 nm of each
replicate was determined. The calibration curve was used to determine the benzoic acid concentration of
each sample.

II. Results and Discussion

Method validation was performed in the analysis of benzoic acid in soft drink sample. The soft
drink sample was first heated in a beaker before analyzing to different tests for method validation. The
sample was heated to expel CO2 from the sample. After, it was filtered to remove the particles and was
cooled to room temperature. Before performing the method validation tests, the UV spectrum of the
blank, benzoic acid, and sample (soft drink) were determined. Based from the overlapped spectrum, no
caffeine interference was detected since the only the spectra of the sample and benzoic acid were found.

Figure 3.1 UV spectrum of Benzoic Acid (left) and Blank Sample (right).

Figure 3.2. UV spectrum of the Sample, Caffeine, and Overlapped Spectrum.

The first test for method validation was selectivity. Selectivity is important so that we can
determine if it can distinguish the analyte from everything else that might be in the sample. The method
determines if the instrument can detect the benzoic acid in the presence of an interference, which is the
caffeine. The λmax of the standard benzoic acid and the caffeine was determined.

Table 3.1. Data on the UV spectra of Blank, Standard Benzoic Acid, and Caffeine.

Solution λmax, nm
Blank ---
Standard Benzoic 230.0
Caffeine Sample 273.0

Table 3.1 Data on the Absorbance of Standard Benzoic Acid Solutions without Caffeine

Concentration, mg/L Average Absorbance

0 0.0000
2 0.2155
 4 0.5540
6 0.6038
8 0.8604
10 1.2601

Table 3.2 Data on the Linear Analysis of Benzoic Acid Calibration Curve without Caffeine.

Parameters Values
slope 0.118352857
Y-intercept -0.009497619
R2 0.9693

Figure 3.3. Calibration curve of Benzoic Acid without Caffeine

Table 3.3 Data on the Absorbance of Standard Benzoic Acid Solutions with Caffeine

Concentration, mg/L Average Absorbance

0 0.0000
2 0.2739
4 0.5518
6 0.8600
8 0.8788
10 1.1049

Table 3.4. Data on the Linear Analysis of Benzoic Acid Calibration Curve.

Parameters Values
Slope 0.109248571
Y-intercept 0.06532381
R2 0.9624
Figure 3.4. Calibration curve of Benzoic Acid with Caffeine

The λmax was set for the standard benzoic acid and the absorbances of the samples were
determined. We can observe that there is a difference between the calibration curve obtained from the
sample without caffeine and sample with caffeine. The linearity of the sample with caffeine decreases,
and the absorbances obtained from the two set of samples were different. We can also observe that the
slope of the two calibrations are different. Interference can cause a bias by increasing or decreasing the
signal attributed to the measurand. Proportional effect is when the size of the effect for a given matrix is
proportional to the signal. While in rotational effect, it changes the slope of the calibration function, but
not its intercept. Other type of interference is the fixed effect, which arises from a signal produced by
interferences present in the test solution (Magnusson and Ornemark, 2014).

Table 3.5. Data on the Absorbance of the Method Blank Sample Solutions of Benzoic Acid.

Replicate (DF=100) Absorbance

1 0.0293
2 0.0239
3 0.0331
4 0.0153
5 0.0181
6 0.0123
7 0.0167
8 0.0283
9 0.0144
10 0.0266
Limit of detection (LOD) is defined as the lowest concentration of the analyte that can be detected by the
method at a specified level of confidence. While limit of quantification is the lowest level at which the
performance is acceptable for a typical application. The standard deviation that is used in calculating the
LOD and LOQ should be a representative of the precision of other test samples. The absorbance of the
ten replicates with DF =100 were obtained. In determining the LOD and LOQ, the calibration curve from
the selectivity assessment was used. However, points were taken off from the graph to improve the
linearity of the calibration curve (specifically concentrations 4, 6, and 8 mg/L). The calculated LOD and
LOQ concentrations for the sample are 0.538801267 and 0.9611813492 mg/L.
Table 3.6. Determination of LOD and LOQ.

Parameters Value
mean 0.0218
Standard deviation 7.307530363 x10-3
YLOD 0.04372259109
CLOD 0.538801267
YLOQ 0.09487530363
CLOQ 0.9611813492
After determining the LOD and LOQ, the trueness of the or bias was determined. According to
Eurachem (2014), trueness is the expression of how close the mean of an infinite number of results
(produced by the method) is to a reference value. In calculating the trueness, the samples were spiked
with 1 mg/L benzoic acid (low spike) and 4 mg/L benzoic acid (high spike). The %Recovery of the two sets
of samples were obtained. The calculated %Recovery for low spike and high spike are 7377.84% and
9282.771745%. Using the t-test for two means, the two set of spikes have significance at 95% confidence
level. There is a bias in the sample.

Table 3.7. Data on the Absorbance of the Test Sample without Spike.

Replicate (DF =100) Absorbance Concentration

1 0.2404 216.2813289
2 0.2343 211.2444142
3 0.2469 221.6485331
4 0.2506 224.7037109
5 0.2365 213.0610064
6 0.2346 211.4921313
Mean 100 216.4051875

Table 3.8. Data on the Absorbance of the Test Sample with Low Spike.

Replicate (DF=100) Absorbance Concentration

1 0.3353 294.6425102
2 0.3239 285.2292598
3 0.3239 285.2292598
4 0.3353 294.6425102
5 0.3311 291.1744706
6 0.361* 315.8636099*
Mean 290.1836021
Concentration with * = OUTLIER

Table 3.9. Data on the Absorbance of the Standards Used in Samples with High Spike.

Concentration, ppm Absorbance

0 0
1 0.074
2 0.174
4 0.348
 6 0.53
8 0.767
10 0.895

Table 3.10. Data on the Absorbance of the Test Samples with High Spike.

Replicate (DF=100) Absorbance Concentration

1 0.686 754.3190719
2 0.656 721.8429025
3 0.673 740.2460652
4 0.676 743.4936821
5 0.661 727.2555974
6 0.65 715.3476687

Table 3.11. Data on the Absorbance of the Test Samples without Spike.

Replicate (DF=100) Absorbance Concentration

1 0.304 340.789182
2 0.318 355.9447277
3 0.317 354.8621887
4 0.31 347.2844159
5 0.316 353.7796497
6 0.307 344.0367989
Mean 349.4494938

Table 3.12. Data on the Determination of %Recovery and Statistical Analysis

Parameter %Recovery
Low-Spike (1mg/L) 7377.84146
High-Spike (4mg/L) 9282.771745
t-test for low spike significant
t-test for high spike significant

The linearity of the method used was also determined. Linearity is important in order to know if the
calibration has errors. In linearity, any curved pattern is not linear and has lack of fit. In measuring linear
calibration, there should be at least six or more calibration standards, the calibration standards should be
evenly spaced over the concentration range of interest, and it should be done at least twice or duplicate
(Thompson et al, 2002). So, for the experiment, 2 calibrations were done with a one week interval. Also,
6 standards were used. For the first week, the calculated R2 is 0.9998, which is linear since its value is close
to 1. For the second week, the calculated R2 is 0.9994. The value of R2 decreased for the second week, but
it is still linear. Linearity can be tested informally by examination of a plot of residuals produced by linear
regression of the responses on the concentrations in an appropriate calibration set. A more complex
method is used for significance testing when the calibration is not linear. When plotting the residuals, a
random distribution of residuals about zero confirms linearity.
Table 3.13.. Net Absorbance of Standard Solutions of Benzoic Acid without HCl (first week)

Concentration, mg/L Net Absorbance

0 0.0000
1 0.0494
2 0.0929
4 0.1810
6 0.2752
8 0.3708
10 0.4568

Table 3.14. Data on the Linear Analysis of the Calibration Curve of Benzoic Acid Standard Solutions without
HCl.

Parameter Value
Slope 0.045744
Y-intercept 0.001158
R2 0.9998

Figure 3.5 .Calibration Curve of Benzoic Acid Standard Solutions without HCl

Figure 3.6. Residual Plot of the Calibration Curve of Benzoic Acid without HCl

Table 3.15. Net Absorbance of Standard Solutions of Benzoic Acid without HCl (second week).
 Concentration, mg/L Absorbance
0 0.0000
1 0.0335
2 0.0669
4 0.1385
6 0.2070
8 0.2787
10 0.3585

Table 3.16. Data on the Linear Analysis of the Calibration Curve of Benzoic Acid Solutions without HCl.

Parameter Value
Slope 0.035613
Y-intercept -0.000299
R2 0.9994

Figure 3. 7. Calibration curve of Benzoic Acid Standard Solutions without HCl (second week).

Lastly, the precision of the method was determined. Precision is the closeness of the data to each other.
Replication is one of the most important factors to consider for us to obtain a reliable estimate of the
precision. In precision, repeatability and reproducibility are the two extreme measurands. Repeatability
is a measure of the variability in results when a measurement is performed by a single analyst using the
same equipment over a short timescale. Repeatability also has the smallest variation in results. On the
other hand, reproducibility is the measure of the variability in results between laboratories. This gives the
largest variations in results ((Magnusson and Ornemark, 2014). In the experiment, the precision was
calculated by having four treatments distributed to two classes. The concentrations of the two treatments
were calculated in the first week and the concentrations for the other two treatments were calculated in
the second week. One-way ANOVA test at 95% significance level was done in order to test the significance
of the calculated mean of the 4 treatments. Based from the result of the One-way ANOVA test, at least
one of the treatment level has a different mean.

Table 3.17 Data on the concentration of samples for precision. (1st week: other 1L)
 Replicate (DF = 50) Absorbance Concentration
1 0.491 271.6119855
2 0.498 275.4008719
3 0.481 266.1992906
4 0.479 265.1167516
5 0.478 264.5754822
6 0.529 292.1802261
Table 3.18. Data on the concentration of samples for precision. (2nd week: 1L)

Replicate (DF = 100) Absorbance Concentration

1 0.304 340.78918
2 0.318 355.94473
3 0.317 354.86219
4 0.31 347.28442
5 0.316 353.77965
6 0.307 344.0368
st
Table 3.19. Data on the concentration of samples for precision. (1 week: class 2L)

Replicate (DF = 100) Absorbance Concentration

1 0.2404 216.2813289
2 0.2343 211.2444142

3 0.2469 221.6485331
4 0.2506 224.7037109
5 0.2365 213.0610064
6 0.2346 211.4921313
Table 3.20. Data on the concentration of samples for precision. (2nd week: class 2L)

Replicate (DF = 100) Absorbance Concentration

1 0.2845 252.6957451
2 0.3043 269.0450748
3 0.2842 252.448028
4 0.288 255.5857781
5 0.2891 256.4940742
6 0.2947 261.1181271

Table 3.21. Data on the One-way ANOVA analysis of the test samples.
 III. Conclusion

Method validation is defined as the confirmation by examination and provision of objective

evidence that the particular requirements for a specific intended use are fulfilled. It is important in order
to provide quality assurance in analyzing. The experiment determines the selectivity, LOD and LOQ,
linearity, trueness and precision of the method in determining the benzoic acid in soft drink sample.

For the selective test, the method was found selective since it can still detect the benzoic acid in
the presence of caffeine. The slope and linearity decreased slightly and also, no interference was found in
the overlapped spectrum.

For the determination of LOD and LOQ, the concentration for LOD was 0.538801267 mg/L while
the concentration obtained for LOQ was 0.9611813492 mg/ L

For the trueness of the method, % Recovery of the low-spiked samples and high-spiked samples
were obtained. The %Recovery for low-spiked sample and high-spiked sample are 7377.84146% and
9282.771745%. Statistical analysis was done using t-test for low-spiked sample and high spiked sample.
The mean for the unspike and low-spike sample was analyzed at 95% confidence level and was found out
to have a significance. Also, the mean for the unspike and high-spike sample was analyzed at 95%
confidence level and was found out to have a significance. There is a bias on the method.
 For the linearity, two replicates were done with a one-week interval. The obtained R2 for the first
week was 0.9998 while the obtained R2 for the second week was 0.9994. A slight decrease was observed
in the second week. But in conclusion, the method is well-calibrated.

Lastly, the precision of the method was determined. The data obtained for two weeks was
compared to the data obtained from other lab class. The data obtained were analyzed using One-Way
ANOVA test and it was found out to have a significance at 95% confidence level. At least one of the
treatment has a different mean.

Possible errors were committed during the experiment that resulted to obtaining bad results.
Errors could be caused by random error, matrix variation effect, method bias, and laboratory effect.

V. Literature Cited

Harris, D. (2007). Quantitative Chemical Analysis 7th Edition. W.H. Freeman and Company. New York

Magnusson B. and Örnemark U. (2014) Eurachem Guide: The Fitness for Purpose of Analytical Methods
– A Laboratory Guide to Method Validation and Related Topics 2nd edition. ISBN 978-91-87461-59-0

Thompson, M., Ellison, S. L. R., Wood, R. (2002). Harmonized Guidelines for Single-Laboratory Validation
of Methods of Analysis (IUPAC Technical Report). Pure Appl. Chem., Vol. 74, No. 5, pp. 835–855, 2002.