Syllabus topics (RTMNU)
- Classification of analytical techniques
- Qualitative and quantitative analysis
- Classical and instrumental methods
- Factors affecting choice of analytical method
- Errors in chemical analysis. Types of errors: Systematic and random, Absolute and relative, Additive and proportional
- Normal distribution of indeterminate errors
- Statistical parameters for data evaluation: Mean, median, average deviation, standard deviation, coefficient of variation, relative standard deviation
- Accuracy and precision of results
- Comparison of data using F and t-test, rejection of data using Q test
- Numerical problems
Classification of analytical techniques
Analytical chemistry deals with finding what is present in a sample (qualitative) and how much is present (quantitative).
- Qualitative analysis: identification of elements, ions, functional groups or compounds (e.g. salt analysis, flame test, spot tests, IR spectrum).
- Quantitative analysis: determination of amount or concentration (e.g. titration of acid, gravimetric estimation of Ba2+, colorimetric estimation of Cu2+).
| Classical (chemical / wet) methods | Instrumental methods |
|---|---|
| Based on chemical reactions: volumetric (titrimetric) and gravimetric analysis | Based on measurement of a physical property: spectroscopy, electroanalysis, chromatography, thermal methods |
| Simple, cheap apparatus; accurate for major constituents | Costly instruments; fast; sensitive for trace amounts; can be automated |
| Slow, need larger samples | Need calibration with standards |
Factors affecting choice of analytical method
- Nature of the information needed (qualitative/quantitative, total or species).
- Amount of sample available and concentration range of the analyte (major, minor, trace).
- Required accuracy and precision.
- Interferences present (matrix).
- Number of samples, time and speed required.
- Cost and availability of equipment and trained staff.
- Whether the method must be non-destructive; safety and environmental issues.
Errors in chemical analysis and types of errors
Error = observed value β true value.
- Systematic (determinate) errors: have a definite cause and direction; can be found and corrected. Types: instrumental (uncalibrated burette, faulty balance), method errors (indicator end point differs from equivalence point, co-precipitation, side reactions), personal errors (colour judgement, parallax), reagent errors (impure chemicals). Reduced by calibration, blank determination, control (standard) samples, independent methods, standard addition.
- Random (indeterminate) errors: small, unpredictable variations in both directions from uncontrollable causes (temperature fluctuation, reading the last digit). They cannot be removed but are treated by statistics; averaging many readings reduces them.
- Absolute error E = x β xt; relative error = (x β xt)/xt Γ 100% (or in ppt).
- Additive (constant) error: same size whatever the sample size (e.g. a constant indicator blank) β its relative effect falls as sample size increases. Proportional error: size increases in proportion to sample size (e.g. an impurity in the sample, wrongly standardised titrant) β relative error constant.
Normal distribution of indeterminate errors
When a measurement is repeated many times, random errors follow the normal (Gaussian) distribution β a symmetrical bell-shaped curve centred on the mean ΞΌ. Small errors are more common than large ones; positive and negative errors are equally likely. About 68.3% of results fall within ΞΌ Β± 1Ο, 95.4% within ΞΌ Β± 2Ο and 99.7% within ΞΌ Β± 3Ο.
Statistical parameters for data evaluation
- Mean xΜ = Ξ£xi/n.
- Median: middle value when results are arranged in order (average of the two middle values for an even number).
- Average deviation dΜ = Ξ£|xi β xΜ|/n.
- Standard deviation s = β[Ξ£(xi β xΜ)2/(n β 1)].
- Relative standard deviation RSD = s/xΜ; coefficient of variation CV = (s/xΜ) Γ 100%.
Accuracy and precision
- Accuracy: closeness of a result to the true (accepted) value; expressed as error. Affected by systematic errors.
- Precision: closeness of repeated results to one another; expressed as standard deviation/CV. Affected by random errors.
- Results can be precise but not accurate (all close together but far from true value β systematic error present). Good analysis needs both.
Comparison of data β F-test and t-test; rejection of data β Q-test
- F-test compares the precision of two sets: F = s12/s22 (larger variance on top). If Fcalc > Ftable, the precisions differ significantly.
- t-test compares a mean with a true value: t = |xΜ β ΞΌ|βn / s; or two means xΜ1, xΜ2: t = (|xΜ1 β xΜ2|/spooled)β[n1n2/(n1 + n2)]. If tcalc > ttable (for n β 1 degrees of freedom at the chosen confidence level), the difference is significant (a systematic error is likely).
- Q-test (Dixon) decides whether a suspect (outlying) result may be rejected: Q = |suspect value β nearest value| / range. If Qcalc > Qcrit the value is rejected. Qcrit (90%): n = 3: 0.94, n = 4: 0.76, n = 5: 0.64, n = 6: 0.56, n = 7: 0.51.
Numerical problems
Example 1. Results (% Cl): 10.08, 10.11, 10.09, 10.10, 10.12. Find mean, median, average deviation, s and CV.
Mean = 50.50/5 = 10.10; median = 10.10; deviations 0.02, 0.01, 0.01, 0.00, 0.02 β dΜ = 0.06/5 = 0.012; Ξ£(dev)2 = 0.0010 β s = β(0.0010/4) = 0.0158; CV = 0.0158/10.10 Γ 100 = 0.16%.
Example 2 (t-test). If the true value for Example 1 is 10.14: t = |10.10 β 10.14| Γ β5 / 0.0158 = 5.66. ttable (4 d.f., 95%) = 2.78. Since 5.66 > 2.78, the difference is significant β a systematic error is present.
Example 3 (Q-test). Normality values: 0.1050, 0.1053, 0.1048, 0.1072. Suspect = 0.1072. Q = (0.1072 β 0.1053)/(0.1072 β 0.1048) = 0.0019/0.0024 = 0.79. Qcrit (n = 4, 90%) = 0.76. Q > Qcrit, so 0.1072 is rejected.
Likely exam questions
- (Long, 8 marks) What are errors? Explain systematic and random errors. How can systematic errors be minimised?
- (Long, 8 marks) Explain the F-test, t-test and Q-test. Replicate results are 20.12, 20.15, 20.11, 20.31. Should 20.31 be rejected at 90% confidence (Qcrit = 0.76)?
- (Short, 4 marks) Distinguish between accuracy and precision.
- (Short, 4 marks) Define standard deviation and coefficient of variation.
- (Short, 4 marks) Distinguish between classical and instrumental methods of analysis.