Figure explanations, laboratory applications and advanced problems
Analytical reliability depends on sampling, calibration, selectivity, blank correction and uncertainty. Precision is repeatability; accuracy is closeness to the accepted value. Detection limit is not the same as reliable quantification.
Detailed interpretation of every chapter diagram
How to interpret it: Analytes partition differently between stationary and mobile phases. Retention time aids identification only with controlled conditions and standards; peak area supports quantification after calibration.
Laboratory connection: Run blanks, standards and quality controls; inspect retention, peak shape, carryover and resolution.
How to interpret it: Within a validated range, absorbance is proportional to concentration, path length and absorptivity. A calibration curve converts response into concentration.
Laboratory connection: Prepare independent standards, run a reagent blank, measure replicates and inspect residuals as well as R².
How to interpret it: IR absorption requires a vibration to change the molecular dipole moment. Functional-group bands are diagnostic; fingerprint-region patterns help distinguish compounds.
Laboratory connection: Collect a background spectrum, control ATR contact or sample thickness and compare with authenticated references.
How to interpret it: Ionisation forms gas-phase ions; an analyser separates them by m/z and a detector records abundance. Fragmentation patterns support structural assignments.
Laboratory connection: Calibrate mass axis, run blanks and internal standards, and distinguish molecular-ion evidence from fragment-ion interpretation.
How to interpret it: Validation assesses selectivity, linearity, range, accuracy, precision, LOD, LOQ, robustness and matrix effects. High R² alone does not prove a valid method.
Laboratory connection: Use replicate standards and quality controls across the working range; assess recovery and within-run and between-run precision.
Additional university-level derivation
Additional derivation: For a linear calibration y=a+bx, least-squares slope b=Σ(xi−x̄)(yi−ȳ)/Σ(xi−x̄)² and intercept a=ȳ−bx̄. For independent uncertainties, if z=xy then (uz/z)²=(ux/x)²+(uy/y)². Report units and the confidence basis of LOD/LOQ.
Advanced numerical exercises with worked answers
Challenge 27.1
Question: Calibration is A=0.0250C+0.0040, C in mg L⁻¹. Sample A=0.379. Find C.
Worked answer: C=(0.379−0.0040)/0.0250=15.0 mg L⁻¹. Apply any dilution factor separately.
Challenge 27.2
Question: A peak has tR=8.40 min and tM=1.20 min. Find retention factor k′.
Worked answer: k′=(tR−tM)/tM=(8.40−1.20)/1.20=6.00.
Challenge 27.3
Question: Two peaks have tR 5.20 and 5.80 min, baseline widths 0.30 and 0.35 min. Find resolution.
Worked answer: Rs=2(5.80−5.20)/(0.30+0.35)=1.85, above the common baseline-separation guide of about 1.5.
Calculation discipline: show the governing equation, substitute values with units, retain appropriate significant figures and state the chemical meaning of the result.
Detailed topic-by-topic lecture development
These expanded notes explain the underlying models, equations, assumptions and applications for each topic in the chapter.
Start with the analytical question: identity, concentration, purity, structure or compliance. Define analyte, matrix, range, selectivity, detection limit, required uncertainty and throughput. Method choice should be fit for purpose; a highly sophisticated instrument is not automatically optimal.
A representative sample is essential. Design sampling for heterogeneity, location, time, sample size and preservation. Homogenisation, digestion, extraction, filtration, dilution and derivatisation can introduce contamination or analyte loss; document each step and maintain chain of custody.
Accuracy describes closeness to a reference value; precision describes agreement among replicate results. Random errors cause scatter, systematic errors cause bias. Use blanks, replicate analyses, certified reference materials, spikes and control charts to monitor performance.
Calibration relates instrument response to analyte amount. External standards are prepared separately; internal standards correct for variability in preparation or injection; standard addition can compensate for matrix effects. Examine residuals and avoid extrapolating beyond the validated range.
LOD is the lowest amount distinguishable from background under defined conditions; LOQ is the lowest amount measured with acceptable performance. Common estimates use blank variability and calibration slope, but method limits must be verified in the actual matrix.
Validation demonstrates fitness for purpose. Selectivity, linearity, range, accuracy, precision, LOD, LOQ, robustness and stability may be assessed. Quality assurance includes training, documented procedures, instrument maintenance, traceability and corrective action.
UV-visible spectroscopy measures electronic absorption. Quantification commonly uses A=εlc. Choose wavelength, cuvette, blank and concentration range carefully. Turbidity, speciation, stray light and spectral overlap can bias absorbance.
IR absorption requires a vibration that changes dipole moment; Raman scattering requires a change in polarisability. The techniques are complementary. Interpret functional-group regions and fingerprint regions while considering resolution, baseline, sample preparation and spectral libraries.
Atomic absorption measures absorption by free atoms; atomic emission measures light emitted by excited atoms or ions. Flame and furnace atomisation, ICP excitation and detector selection influence sensitivity. Correct for spectral, chemical and matrix interferences.
NMR chemical shifts, integration, multiplicity and coupling constants provide structural evidence. Quantitative NMR requires a suitable standard, calibrated pulse, sufficient relaxation delay and non-overlapping signals. Solvent, temperature and exchange can affect spectra.
Chromatography separates analytes through differing distributions between mobile and stationary phases. Retention, selectivity and efficiency control resolution. Optimise stationary-phase chemistry, mobile-phase composition, temperature and flow rather than relying only on longer run time.
GC suits volatile, thermally stable analytes or compounds made volatile by derivatisation. HPLC and UHPLC suit less volatile or thermally labile species. Detector choice (FID, UV, fluorescence, conductivity or MS) depends on analyte properties and matrix.
Mass spectrometry ionises molecules and separates ions by mass-to-charge ratio. EI, ESI and MALDI create ions through different mechanisms. Accurate mass, isotope patterns and fragmentation help identify compounds, but a spectrum alone may not uniquely identify an unknown.
Potentiometry measures potential at near-zero current; voltammetry records current as potential varies; amperometry measures current at fixed potential; conductometry measures ionic conductance. TGA tracks mass change with temperature; DSC tracks heat flow and transitions.
Chemometrics applies multivariate statistics to chemical data. PCA explores patterns and outliers; regression predicts properties from spectra or sensor arrays. Prevent overfitting through proper train/test separation, cross-validation and external validation; report preprocessing and scaling.
Detailed university-level treatment, derivations and applications
Analytical chemistry: calibration, uncertainty and separation theory
1. Calibration by least squares
For a linear response y=mx+b, least-squares regression selects m and b to minimise the sum of squared residuals. The slope converts signal to concentration and the intercept captures background response. Inspect residuals for curvature and heteroscedasticity; a high R² alone does not prove a valid calibration. Use weighting when variance changes systematically across the range.
2. Detection limit and quantification limit
A common estimate is LOD≈3.3σ/m and LOQ≈10σ/m, where σ is an appropriate standard deviation of blank or low-level response and m is calibration slope. These are estimates, not universal physical constants. Validate limits in the actual matrix with replicate low-level samples and a stated probability of false positives/negatives.
3. Chromatographic efficiency and resolution
The retention factor is k′=(tR−tM)/tM. Column efficiency can be estimated from N=16(tR/wb)² using baseline peak width or N=5.54(tR/w1/2)² using half-height width. Resolution is Rs=2(tR,2−tR,1)/(w1+w2). Resolution improves with efficiency, selectivity and suitable retention; very long retention alone does not guarantee good separation.
4. Beer–Lambert and standard addition
For A=εlc, concentration follows from the calibration slope when path length and chemical form are controlled. In standard addition, known analyte increments are added to equal sample aliquots. Extrapolating the signal line to zero response estimates the negative original analyte concentration on the x-axis, compensating for a constant matrix effect under linear response assumptions.
5. Measurement uncertainty
For independent uncertainty contributions, combine standard uncertainties by root-sum-of-squares after applying sensitivity coefficients. For a result y=f(x1,…,xn), u²(y)=Σ(∂f/∂xi)²u²(xi) when inputs are independent. Correlated inputs require covariance terms. Report expanded uncertainty U=ku with the coverage factor and interpretation stated.
Fully worked numerical exercises
Worked numerical problem 1: Concentration from calibration line
Problem: A calibration is signal=0.0250C+0.0040, with C in mg L⁻¹. An unknown gives signal 0.254.
Solution: C=(0.254−0.0040)/0.0250=10.0 mg L⁻¹. This is valid only if the unknown response lies within the validated calibration range and matrix effects are controlled.
Worked numerical problem 2: Estimate LOD and LOQ
Problem: A calibration slope is 0.080 signal units per mg L⁻¹ and blank SD is 0.0020 signal units.
Solution: LOD≈3.3×0.0020/0.080=0.0825 mg L⁻¹. LOQ≈10×0.0020/0.080=0.250 mg L⁻¹. Confirm these estimates experimentally in the sample matrix.
Worked numerical problem 3: Chromatographic retention factor
Problem: tR=8.0 min and tM=1.5 min. Calculate k′.
Solution: k′=(8.0−1.5)/1.5=4.33. The analyte is retained beyond the mobile-phase hold-up time.
Worked numerical problem 4: Chromatographic resolution
Problem: Two peaks have tR1=5.20 min, tR2=5.80 min, and baseline widths 0.30 and 0.40 min.
Solution: Rs=2(5.80−5.20)/(0.30+0.40)=1.71. This is generally close to baseline separation for approximately Gaussian peaks, though matrix and peak shape matter.
Worked numerical problem 5: Uncertainty propagation for a product
Problem: A=2.00±0.02 and B=3.00±0.03 are independent; calculate AB and standard uncertainty.
Solution: y=6.00. Relative uncertainty u(y)/y=√[(0.02/2.00)²+(0.03/3.00)²]=0.0141. Thus u(y)=0.0849; report 6.00±0.08 in consistent units.
Worked numerical problem 6: Standard addition estimate
Problem: Signal is y=0.120x+0.360 where x is added analyte concentration in mg L⁻¹.
Solution: The x-intercept is −b/m=−3.00 mg L⁻¹, so the original analyte concentration is estimated as 3.00 mg L⁻¹ after accounting for dilution. The method assumes linear response and equal matrix composition across aliquots.
Short-answer revision with model answers
Model answer 1
Question: Why can a high R² calibration still be poor?
Answer: A high R² can coexist with curvature, outliers, heteroscedasticity, biased standards or extrapolation. Inspect residuals, range, back-calculated standards and independent quality controls.
Model answer 2
Question: Distinguish precision from accuracy.
Answer: Precision is agreement among repeated measurements. Accuracy is closeness to the accepted reference value. Replicates can be precise but biased.
Model answer 3
Question: Why is representative sampling part of analytical quality?
Answer: A perfectly calibrated instrument cannot correct a sample that does not represent the material or population of interest. Sampling bias propagates into every later result.
Model answer 4
Question: When is standard addition useful?
Answer: It is useful when matrix effects alter analyte response and matrix-matched standards are difficult to prepare. It assumes the added and native analyte behave equivalently and response is linear.
Model answer 5
Question: What does chromatographic resolution depend on?
Answer: Resolution depends on column efficiency, selectivity between analytes and retention. Improve stationary/mobile phase chemistry and peak efficiency rather than relying only on longer retention.