Spectrophotometry: the Beer–Lambert law
A colored solution absorbs part of the light that crosses it, and the more concentrated it is, the more it absorbs. That is the basis of spectrophotometry, the most used method to measure concentrations in the laboratory. This topic explains the Beer–Lambert law, transmittance, calibration curves and how a mixture of several species is analyzed.
Transmittance and absorbance
If an intensity arrives and leaves, the transmittance is (or %T = 100·T). Each thin layer of solution absorbs the same fraction of the light reaching it, so the intensity falls exponentially and what grows in proportion to concentration is the logarithm, the absorbance:
is the molar absorptivity (L·mol⁻¹·cm⁻¹), specific to each species and each wavelength; is the path length (almost always 1 cm) and the molar concentration. A = 1 lets 10% of the light through and A = 2 lets 1%: that is why it works best with absorbances between 0.1 and 1, where the relative reading error is small.
Calibration curve
In practice is rarely known in advance: standards of known concentration are prepared, their absorbance is measured at the wavelength of maximum absorption and A is plotted against c. The slope is ; the concentration of a sample is read on the line. With a single standard, .
Mixtures: additivity
If the sample has several species that do not react with each other, their absorbances add up. Measuring at as many wavelengths as species (preferably near each one's maximum) leaves a linear system with the concentrations as unknowns:
The system is well conditioned when each wavelength «sees» mostly one species; if the spectra are very similar, a small reading error changes the concentrations a lot. With more wavelengths than species it is solved by least squares, which averages the errors, and the residual tells how well the readings fit.
Common mistakes
- Using %T instead of A in , or taking without converting to a fraction.
- Mixing units: ε in L·mol⁻¹·cm⁻¹ requires b in cm and c in mol/L.
- Extrapolating the calibration curve beyond the last standard.
- Using the ε of one wavelength for readings taken at another.
- Forgetting to subtract the blank (the cell with only the solvent).
Worked examples
Example 1From %T to concentration
Solution
- → .
- M = 141.3 µM.
Solve it in Simuveliq
- In Chemistry → Spectrophotometry (Beer–Lambert), the Beer's law: A = ε b c card with the reading in %T = 45, ε = 2455, b = 1 cm and c blank (the button sets it): it gives c = 1.413×10⁻⁴ M and both steps (A = −log T, then c = A/εb). The Absorbance ↔ transmittance card does just the conversion.
Example 2Mixture of two species
Solution
- System: and .
- Determinant: . By Cramer: M.
- M.
- Check at 510 nm: Co contributes and Ni ; they add up to 0.446.
Solve it in Simuveliq
- In the Mixture of several species at several wavelengths card, «Or load an example» → «Co²⁺ and Ni²⁺ at 510 and 656 nm (2 species)». The Result gives both concentrations; the Setup writes both equations, and the bar chart shows each species' contribution to each absorbance.
Example 3Readings in %T
Solution
- First to absorbance: and .
- System: and , with determinant 866 100.
- M = 1.157 mM of dichromate and M = 0.2795 mM of permanganate. The ε are small (hundreds, not tens of thousands), so millimolar concentrations are needed to read useful absorbances.
Solve it in Simuveliq
- Example «Dichromate and permanganate, readings in %T»: in step 1 the instrument gives %T and the concentrations are in mM; the tool converts each reading to A before solving.
Example 4ε from standards
Solution
- Each standard gives its ε: , that is, A: 9800 and 1200; B: 1500 and 8500.
- System: and → M and M.
Solve it in Simuveliq
- Example «ε from 1.0×10⁻⁴ M standards»: in step 2 each species has «I get it from a standard of that species alone», and in step 3 the absorbances of the standards are typed instead of ε.
Example 5More wavelengths than species
Solution
- Four equations and two unknowns: there is no exact solution, because the readings have errors. Least squares finds the that minimize the sum of the squared differences between measured and computed A; one solves .
- Result: µM and µM. The recomputed A differ from the measured ones by less than 0.003 (root-mean-square error 0.00157): the readings are consistent.
Solve it in Simuveliq
- Example «More wavelengths than species (least squares)»: the result says so («4 wavelengths for 2 unknowns: least-squares fit») and the table compares, for each λ, the measured A with A = b Σ εc and their difference.