Analytical Chemistry

Beer-Lambert Law

A = εcl — absorbance is linear in concentration and path length; foundation of UV-Vis quantification

The Beer-Lambert law states that absorbance A equals molar absorptivity ε times concentration c times path length l, written A = εcl. Discovered piecewise by Pierre Bouguer in 1729 (path-length dependence), reformulated by Johann Lambert in 1760, and extended to concentration by August Beer in 1852, it is the quantitative foundation of UV-Vis, fluorescence excitation, and infrared spectroscopy. Linearity holds for dilute, non-aggregating, non-fluorescent samples below roughly 0.01 M; above that, chemical and instrumental deviations bend the calibration curve.

  • EquationA = εcl = -log10(I/I0)
  • ε unitsM⁻¹·cm⁻¹
  • Linear rangeA < 1, c < 0.01 M
  • DiscoveredBouguer 1729, Lambert 1760, Beer 1852
  • Best A range0.1 to 1.0 (10–80% T)
  • Used inUV-Vis, IR, AA, fluorescence excitation

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

Why Beer-Lambert matters

  • Universal quantitative tool. Every UV-Vis spectrophotometer in every analytical lab — from undergraduate teaching to FDA-regulated pharmaceutical QC — uses A = εcl as its working equation. Concentrations from millimolar drug solutions down to 10⁻⁷ M dye assays are read off calibration curves built on this single line.
  • Wavelength-resolved selectivity. Because ε(λ) is structure-specific, choosing the analytical wavelength selects the analyte. Hemoglobin at 415 nm (Soret band, ε ≈ 125,000) is read directly in plasma without separation; the same cuvette at 540 nm gives oxyhemoglobin alone.
  • Path length is engineering-controllable. Standard quartz cuvettes are 1.000 ± 0.005 cm; flow cells extend to 10 cm for trace nitrate in water (LOD ≈ 1 µg/L NO3-N) or shrink to 1 mm for protein UV absorbance at A280 in concentrated stocks.
  • Provides the language for ε tabulation. Reference handbooks and online databases (NIST WebBook, PubChem) report ε(λ) at named maxima; any new lab can recreate a literature method with no calibration of its own provided ε is known and Beer-Lambert holds.
  • Underpins atmospheric and astronomical optics. Stellar extinction, atmospheric ozone column densities (Dobson units), and oceanographic chlorophyll-a satellite retrievals all invoke Beer-Lambert at long paths and low concentrations where deviations are negligible.
  • Differential measurements amplify sensitivity. A two-cuvette photometer compares (sample + reagent) versus (reagent only); at A < 0.001 baseline drift dominates, and dual-beam optics push detection floors to 10⁻⁹ M for chromophores with ε > 10⁵.
  • Foundational for modern bioassays. ELISA plate readers measure 96- to 1536-well TMB or pNPP color development as A450 or A405; the law converts absorbance into ng/mL antigen via a sigmoidal four-parameter logistic fit anchored on the linear regime.

Common misconceptions

  • Beer's law works at all concentrations. Above ~0.01 M, electrostatic and hydrogen-bonding interactions between solutes alter the local field around chromophores and ε drifts; aggregation forms dimers and trimers with their own ε. Always dilute into the linear regime, never extrapolate a calibration curve past its highest verified standard.
  • Transmittance is linear in concentration. Transmittance T = 10⁻εcl is exponential, which is why doubling concentration drops T from 50% to 25% (not to 0%). Plot absorbance, not transmittance, against concentration; only A is linear in c.
  • ε is a property of the molecule alone. ε depends on solvent (polarity shifts λ_max and band shape — the solvatochromic effect), temperature, and pH for ionizable groups. Indicators are extreme cases: phenolphthalein has ε ≈ 0 in acid and ε ≈ 26,000 at 553 nm above pH 10.
  • Stray light only affects high concentrations. Stray light caps maximum measurable absorbance: at 0.01% stray light A_max ≈ 4, at 1% stray light A_max ≈ 2. Modern double-monochromator instruments achieve 0.0008%, single-monochromator instruments hover near 0.05%. Verify by measuring absorbance of a holmium oxide reference well past the published peak.
  • Background subtraction fixes everything. Subtracting blank absorbance handles cuvette/solvent absorbance but does nothing for matrix-induced changes to ε of the analyte. Method of standard additions (spike known amounts into matrix-matched samples) is the rigorous fix when the matrix shifts ε.
  • Polychromatic light gives the same A as monochromatic. If ε varies sharply across the spectral bandwidth of the monochromator, the measured A is a transmittance-weighted average that is always less than the monochromatic A — Beer's law fails on a sloped peak even at infinitely low c. Use bandpass ≤ 1/10 of the band's FWHM, typically 1 nm for a 10 nm band.

Derivation from photon survival

Consider a parallel beam of intensity I traveling through a slab of thickness dl containing N absorbing molecules per cm³, each with photon-capture cross-section σ (cm²). The probability that a photon is absorbed in the slab is σ·N·dl, and the differential intensity drop is dI = -σN·I·dl. Integrating from 0 to l: I(l) = I0·exp(-σNl). Converting natural to decadic logarithm and identifying the decadic absorption coefficient as σN·log10(e) gives -log10(I/I0) = (σN/2.303)·l. Substituting molar concentration c (mol/L) for number density via N = c·N_A·10⁻³, and lumping molecule-specific constants into ε = σ·N_A·10⁻³/2.303, yields A = εcl.

The form of the law assumes independence of absorbers (no cooperativity), a single chromophore species (no equilibria during the measurement), monochromatic light (single wavelength), and a homogeneous sample (no scattering). When any assumption breaks the law deviates: aggregating dyes show negative deviations (lower A than expected); polychromatic light gives negative deviations on sloped peaks; scattering particles add an apparent absorbance that scales as λ⁻⁴ and rises toward the UV.

Solving for ε from a measured spectrum requires the calibration design: standards traceable to a reference material, cuvettes matched to ±1% in path length (or single-cuvette with rinse-in-place), and an internal blank. Reporting ε at a non-maximum is permissible but error-prone because small monochromator wavelength miscalibrations translate directly into apparent ε errors. The community convention is to report ε at λ_max ± 0.5 nm.

Beer-Lambert deviations: chemical versus instrumental

Deviation typeCauseSign of errorConcentration regimeFixTypical magnitude
Aggregation/dimerizationSolute-solute associationNegative (A under-predicted)c > 10⁻³ MDilute, change solvent5 to 50%
Acid-base equilibriaTwo species with different εEither, pH-dependentNear pKaBuffer at pH ≫ or ≪ pKaUp to 100%
Refractive index changeLocal-field correctionEitherc > 10⁻² MBracket with matrix-matched standards1 to 5%
Stray lightNon-monochromatic photons reach detectorNegative at high AA > 1.5Dilute, use double-monochromator5 to 50% above A = 2
Polychromatic lightBandpass > band widthNegative on sloped peaksAnyReduce slit width1 to 20%
Light scatteringParticles, turbidityPositive (false A)Any with particulatesFilter 0.22 µm, blank with scatterer10 to 100%
Fluorescence reabsorptionSample re-emits into beamNegativeA > 0.1 of fluorophoresUse perpendicular geometry2 to 10%

Famous applications and worked numbers

  • Protein A280 quantification. Tryptophan ε280 ≈ 5,690 and tyrosine ε280 ≈ 1,280 M⁻¹·cm⁻¹ give a sequence-derivable ε for any protein. NanoDrop and similar microvolume spectrophotometers measure 1 µL at 1 mm path length and report mg/mL by Beer-Lambert; an IgG with ε280 = 1.4 (mL/mg/cm) at A280 = 0.7 is therefore 0.5 mg/mL.
  • Hemoglobin in clinical hematology. Cyanmethemoglobin (Drabkin's method) absorbs at 540 nm with ε = 11,000 M⁻¹·cm⁻¹; a normal adult male reads A540 ≈ 0.73 in a 1 cm cuvette diluted 1:251, corresponding to roughly 15 g/dL hemoglobin. Standardized by ICSH since 1966 and still the WHO reference method.
  • Phosphate in environmental water. Molybdenum-blue chemistry yields ε ≈ 26,000 at 880 nm; with a 5 cm long-path flow cell this gives an LOD around 0.5 µg/L PO4-P, sufficient for oligotrophic lake monitoring.
  • DNA A260/A280 purity ratios. Pure dsDNA has ε260 = 6,600 M⁻¹·cm⁻¹ per nucleotide and an A260/A280 ratio near 1.8 (1.0 OD ≈ 50 µg/mL); protein contamination drops the ratio toward 1.5, while phenol contamination from extraction shifts A260/A230 below 2.0. The ratio is a Beer-Lambert difference at two wavelengths and is reported on every NanoDrop trace.
  • Atmospheric ozone — Dobson units. Dobson spectrophotometers measure ratios of solar UV at 305 and 325 nm, where ozone has ε differing by a factor of ten. Beer-Lambert with the known O3 absorption cross-section (Hearn 1961, refined by Daumont 1992) inverts to total column ozone — the original 1985 Antarctic ozone hole detection used a 1957-design Dobson at Halley Bay.

Frequently asked questions

What does A = εcl mean physically?

Each absorbing molecule presents an effective cross-section σ to a photon of the right wavelength. The probability that a photon survives a slab of thickness dl with number density N is exp(-σN dl). Integrating over the cuvette path gives transmittance T = exp(-σNl), and converting natural log to base-10 log lumps σN into the molar absorptivity ε times concentration c. So A = -log10(I/I0) = εcl says: doubling either how many absorbers a photon meets per unit length or how much length it traverses doubles the absorbance (not the transmittance). Absorbance is a dimensionless logarithm; ε carries units of M⁻¹·cm⁻¹, c is molar, and l is in cm by convention.

Why does linearity break down above about 0.01 M?

Three effects compete. First, solute molecules begin to interact electrostatically; refractive index changes alter the effective ε. Second, dimerization or aggregation creates a new species with different ε, so the apparent ε is a mole-fraction-weighted average that drifts with c. Third, on the instrument side, stray light (light reaching the detector that bypassed the sample) caps the maximum measurable absorbance — at A = 2 with 0.1% stray light, the apparent absorbance is already 5% low. Practical UV-Vis work keeps absorbance between 0.1 and 1.0 (transmittance 79% to 10%), where signal-to-noise is best and deviations are below 1%.

How big is ε for typical chromophores?

Allowed π→π* transitions in conjugated aromatics are around ε ≈ 5,000 M⁻¹·cm⁻¹ at the maximum, e.g. benzene 256 nm peak is roughly 200 M⁻¹·cm⁻¹ (forbidden), naphthalene 220 nm is about 100,000. Charge-transfer bands in transition metal complexes such as MnO4⁻ at 525 nm reach ε ≈ 2,500. Dyes engineered for visibility — fluorescein, rhodamine — climb to ε ≈ 80,000 to 100,000. Symmetry-forbidden d-d transitions of first-row transition metals like [Cu(H2O)6]²⁺ blue color are weak, ε ≈ 10. Water itself absorbs negligibly across 200 to 700 nm but jumps near 200 nm where ε ≈ 10⁴ M⁻¹·cm⁻¹ from σ→σ* cuts off practical UV measurement at this wavelength.

What is the difference between absorbance and absorbance per unit length?

Absorbance A is dimensionless (log10 of an intensity ratio). The per-length quantity is the absorption coefficient α = εc with units cm⁻¹, sometimes called the linear extinction coefficient. In atmospheric and biological optics where path lengths vary, α is more convenient and Beer-Lambert reads I = I0 exp(-αl) with the natural-log α = ε·c·ln10 ≈ 2.303 εc. Confusingly, two ε constants coexist in the literature: ε10 (decadic, our convention) and εe (Napierian); they differ by a factor of ln10 ≈ 2.303. Always check journal or instrument convention before borrowing values.

How is a UV-Vis calibration curve constructed?

Prepare 5 to 7 standards spanning the expected range, ideally producing absorbances from 0.1 to 1.0 at the analytical wavelength. Run them in matched cuvettes after a blank subtraction (solvent-only baseline). Plot A vs c and fit a least-squares line; intercept should be within 3σ of zero, slope is ε·l. R² typically exceeds 0.999 for a well-behaved analyte. Limit of detection LOD = 3.3 σ_blank / slope, limit of quantification LOQ = 10 σ_blank / slope. For non-linear regimes use a quadratic fit but flag the result as semi-quantitative; better practice is to dilute into the linear regime.

Does Beer-Lambert apply to fluorescence and IR too?

It governs the absorption (excitation) step in both. Fluorescence intensity is roughly proportional to εcl·Φ_F·I0 only when εcl < 0.05 — above that, inner filter effects make a fraction of incident photons absorbed by molecules out of the detection volume, so fluorescence drops below linearity. In IR, Beer-Lambert holds for dilute solutions, but condensed-phase samples often show bandshape distortion from refractive-index dispersion (Kramers-Kronig coupling) that shifts apparent ε up or down by 5 to 30%. Quantitative IR therefore reports peak height ratios using internal standards rather than absolute ε values.