Spectrophotometry & Beer’s Law · Guide · glidepathscience.org
GuideSpectrophotometry & Beer’s Law
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round. Part 1 prepares the Intro track and Part 2 the Advanced track. Each section ends with a worked example that is not one of the sim’s own problems.
Part 1 · The essentials
Math you’ll need: algebra, powers of ten, and the base-10 logarithm ($\log_{10}100 = 2$, $\log_{10}0.1 = −1$); one optional box uses a derivative.
Light in, light out
A spectrophotometer sends light of one color through a small transparent cell, the cuvette, and a detector measures what comes out. A lamp makes white light; a monochromator (a prism or a grating and a slit) picks out a narrow band of wavelengths $λ$. With only solvent in the cell the detector reads $I_0$; with a colored solution it reads less, $I$, because molecules of the solute absorb some of the photons.
The fraction that gets through is the transmittance, and its logarithm is the absorbance:
$T = I/I_0, \qquad A = −\log_{10}T = \log_{10}(I_0/I)$
$T$ is often given as a percent (%T). A sample passing 10 % has $A$ = 1, one passing 1 % has $A$ = 2, and a clear sample has $A$ = 0. Absorbance has no units.
Why use a logarithm at all? Because of how light is absorbed, as the next sections show.
The blank
The glass walls reflect a few percent of the light, and the solvent may absorb a little. Those losses have nothing to do with the solute, so the instrument is first set to read 100 %T with a blank: the same cuvette filled with the solvent alone. Losses multiply transmittances, so they add absorbances; zeroing on the blank subtracts them exactly, and the reading is the solute’s alone. Skip it and every absorbance reads too high by the cell’s own $−\log_{10}T_{cell}$. A single-beam instrument must be re-zeroed whenever the wavelength changes, because the lamp’s brightness and the detector’s response change with $λ$.
Color and the wavelength to use
A solution’s color is the light it does not absorb. A blue dye absorbs orange-red light and passes blue and green; a red solution absorbs green; a yellow one absorbs violet and blue. Pairs like these sit opposite each other on a color wheel, so they are called complementary.
Plotting $A$ against $λ$ gives the solution’s absorption spectrum, usually a broad band tens of nanometers wide. Measurements are made at its peak, $λ_{max}$, for two reasons: the absorbance per molecule is largest there, so the method is most sensitive, and the band is flattest there, so a small error in the wavelength setting barely changes the reading.
Beer’s law
Picture the beam crossing the solution slice by slice. Each thin slice holds the same number of molecules, so each absorbs the same fraction of the light that reaches it. Two equal slices pass $T × T$; doubling the concentration (twice the molecules in each slice) or doubling the path (twice the slices) therefore squares the transmittance. Taking the logarithm turns that product into a sum, so the absorbance doubles. That is the reason for the logarithm: absorbance is proportional to the number of molecules in the beam,
$A = εbc$
Here $b$ is the path length in cm (a standard cuvette is 1.00 cm), $c$ the concentration in mol/L, and $ε$ the molar absorptivity in L·mol⁻¹·cm⁻¹: the molecule’s own absorbing strength at that wavelength. Strong dyes have $ε$ near 10⁵; pale ions such as Cu²⁺ manage about ten.
Standards, the calibration line and the unknown
In practice nobody trusts a value of $ε$ from a book: the instrument, the slit and the solvent all matter. Instead a chemist makes standards of known concentration, usually by diluting a stock solution. The moles of solute are the same before and after dilution, so
$c_1V_1 = c_2V_2$
The standards are measured at $λ_{max}$, and $A$ is plotted against $c$. The best straight line through them, found by least squares, is the calibration curve; its slope is $εb$ and its intercept should be near zero if the blank was good. An unknown of the same solute is then measured, and its concentration read from the line:
$c = (A − \text{intercept})/\text{slope}$
The working range
Beer’s law is only as good as the measurement. Above an absorbance of about 1.5, less than 3 % of the light reaches the detector, and stray light and noise swamp it: the measured curve bends over and flattens. Below about 0.05 the sample is barely different from the blank. So chemists work at $A$ ≈ 0.1 to 1, and keep the unknown inside the range of the standards. A sample that is too dark is diluted by a known factor, measured, and multiplied back; a shorter cell also works.
Part 2 · Going further
Math you’ll need: exponentials and natural logarithms, a first-order differential equation, two linear equations in two unknowns, and a sample standard deviation.
Beer–Lambert from the molecules
In a slice of thickness $dx$ containing $n$ absorbing molecules per cm³, each presents an effective target area, its absorption cross section $σ$, so the slice removes the fraction $σn\,dx$: $dI/dx = −σnI$ and $I = I_0e^{−σnb}$. Converting to base 10 and to moles ($n = cN_A/1000$ per cm³ for $c$ in mol/L),
$ε = \dfrac{σN_A}{1000\,\ln 10}, \qquad T(x/2) = \sqrt{T(x)}$
Each half of a path passes the square root of the whole path’s transmittance. A molecule’s $σ$ is a few Ų for a strong dye, comparable to the size of its light-absorbing part. In solution each band is broadened by the molecule’s surroundings into a smooth shape tens of nanometers wide; for a Gaussian band of standard deviation $s$ the full width at half maximum is $2\sqrt{2\ln 2}\,s ≈ 2.355\,s$.
Mixtures
Molecules absorb independently, so absorbances add: $A(λ) = b\sum_i ε_i(λ)c_i$. With two absorbers, two wavelengths give two equations in two unknowns, solved by elimination or Cramer’s rule:
$c_1 = \dfrac{A_1ε_{22} − A_2ε_{21}}{D}, \quad c_2 = \dfrac{ε_{11}A_2 − ε_{12}A_1}{D}, \quad D = ε_{11}ε_{22} − ε_{21}ε_{12}$
(with $ε_{ij}$ the absorptivity of species $j$ at wavelength $i$ and $b$ = 1 cm). The answer is reliable when each wavelength is dominated by a different species; when the two spectra have nearly the same shape, $D$ is small and small errors in $A$ become large errors in $c$.
Why the line bends
Stray light. A fraction $s$ of light outside the selected band reaches the detector anyway, so it reads $T′ = (T + s)/(1 + s)$. When $T$ is large this hardly matters; when $T$ approaches $s$ it dominates, and the measured absorbance levels off near $−\log_{10}s$.
Polychromatic light. A slit passes a band of wavelengths and the detector averages the transmittance over it. On a sloping part of a band, the average of $10^{−A}$ is larger than $10^{−\bar A}$, so the meter reads low, and more so at high concentration: the line bends down. A narrow slit at $λ_{max}$ keeps this small.
Chemistry. If the solute exists in two forms in equilibrium, an acid–base indicator for example, their proportions shift with conditions and $A$ at most wavelengths is not proportional to the total concentration. Where the two forms have equal $ε$, the isosbestic point, every spectrum crosses, and there $A$ measures the total whatever the split.
How well is the unknown known?
The least-squares line through $n$ standards has slope $m = \sum(c_i−\bar c)(A_i−\bar A)/\sum(c_i−\bar c)^2$ and intercept $a = \bar A − m\bar c$; the scatter of the points about it gives $s_y$. An unknown measured $k$ times with mean $\bar A_0$ has concentration $c_0 = (\bar A_0 − a)/m$ with standard uncertainty
$s_{c_0} = \dfrac{s_y}{|m|}\sqrt{\dfrac1k + \dfrac1n + \dfrac{(\bar A_0 − \bar A)^2}{m^2\sum(c_i − \bar c)^2}}$
More readings, more standards, and an unknown near the middle of the standards all shrink it. The limit of detection is the concentration whose signal stands clear of the blank’s noise: with $s_{blank}$ the standard deviation of repeated blanks, LOD = $3s_{blank}/m$ and the limit of quantitation LOQ = $10s_{blank}/m$ (common conventions). Finally, if the instrument’s error is a fixed amount of transmittance, the relative error in $c$ is proportional to $1/(T|\ln T|)$, which is smallest at $T = 1/e$, $A$ = 0.434, and stays near its minimum from about 0.2 to 0.8.
Standard addition
When the sample’s own matrix (other dissolved substances, turbidity, pH) changes $ε$, a calibration made in pure water misleads. In standard addition, equal portions $V_x$ of the sample go into flasks of volume $V_t$ with increasing volumes of a standard. Plotting $A$ against the added concentration in the flask gives a line whose extrapolation to $A$ = 0 lies at minus the sample’s contribution, $a/m$; the sample itself is $(a/m)\,V_t/V_x$. Because every flask contains the matrix, the slope already includes its effect.
Reference
Common mistakes
Part 1
- Halving $T$ when $c$ doubles: $T$ squares; $A$ doubles.
- Using %T in the logarithm: $A = −\log_{10}T$ with $T$ as a fraction.
- Using the natural log for absorbance: $A$ is defined with $\log_{10}$.
- Naming the color seen as the color absorbed: they are complements.
- Skipping the blank, or not re-zeroing at a new wavelength.
- Leaving µM unconverted in $ε = A/(bc)$: $c$ must be in mol/L.
- Adding the pipetted volume to the flask’s: the flask is filled to its mark.
- Reading an unknown far outside the standards, where the line is not known.
Part 2
- Leaving out $\ln 10$ between $ε$ and $σ$.
- Solving one wavelength at a time for a mixture whose bands overlap.
- Averaging absorbances for a wide slit: the detector averages transmittances.
- Using the LOQ’s 10 for the LOD’s 3, or the reverse.
- Forgetting the dilution factor $V_t/V_x$ in standard addition.
The equation sheet
| Part 1 | |
|---|---|
| transmittance | $T = I/I_0$ |
| absorbance | $A = −\log_{10}T$ |
| Beer’s law | $A = εbc$ |
| dilution | $c_1V_1 = c_2V_2$ |
| the unknown | $c = (A − a)/m$ |
| Part 2 | |
|---|---|
| attenuation | $I = I_0e^{−αx}$ |
| cross section | $σ = ε\ln 10 × 1000/N_A$ |
| mixtures | $A = b\sum ε_ic_i$ |
| stray light | $T′ = (T + s)/(1 + s)$ |
| detection | LOD $= 3s/m$, LOQ $= 10s/m$ |
| standard addition | $c_{sample} = (a/m)V_t/V_x$ |
Spectrophotometry & Beer’s Law v1.1 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org