Press play to scan from 190 to 1100 nm and watch which lamp is in use. No single lamp covers the whole UV-visible range. A deuterium arc lamp gives a continuum in the ultraviolet; its sharp emission lines at 486.0 and 656.1 nm are used to check the wavelength scale (European Pharmacopoeia). A tungsten-halogen lamp behaves like a hot blackbody: at 2870 K (the textbook filament temperature, Skoog) Wien’s law puts its peak at 1010 nm, in the infrared, and only about 9% of its output falls between 400 and 750 nm in this model. Move the temperature slider to see the peak shift. The deuterium continuum is drawn with an illustrative shape; only its line positions are quantitative.
Choose a grating or a prism. The close-up on the right shows how the light is actually separated. At a grating, the incoming crests reach neighbouring grooves at different moments, and each groove sends out a circular wavelet. In only one direction do the wavelets from all the grooves arrive in step: where the extra path between neighbouring grooves, d sin α + d sin β (drawn in red), equals exactly one wavelength. There they build a new plane wave (bold lines); everywhere else they cancel. Each colour has a different wavelength, so each finds its own direction. At a prism, light slows to c/n inside the glass; its crests crowd together and the wave bends at the surface, and because n is larger for shorter wavelengths, violet and UV bend more than red. Wavelengths and the groove spacing are drawn to the same scale. A reflection grating sends each wavelength off at its own angle, following mλ = d(sin α + sin β), where d is the groove spacing, α the angle of incidence and β the angle of diffraction. More grooves per millimetre spread the colours further (angular dispersion dβ/dλ = m/(d cos β)). The same angle is reached by 600 nm in first order and by 300 nm in second order, so instruments need an order-sorting filter. Worked example checked by the model: 1200 grooves/mm, α = 25°, 500 nm gives β ≈ 10.2° and, with a 200 mm focal length, a reciprocal linear dispersion of 4.1 nm/mm. The rays land on a white card: the visible ones show their colours, while ultraviolet (below 400 nm) and infrared light land where the eye sees nothing, drawn dashed. A prism works by refraction instead: the refractive index rises toward the ultraviolet (Sellmeier equation; fused silica from Malitson 1965, N-BK7 from Schott), so a prism spreads the UV widely and squeezes the red together, unlike the nearly even spread of a grating. Ordinary BK7 glass absorbs the UV, so UV instruments use fused silica or a grating. Press play to follow one wavelength from 200 to 800 nm.
A Czerny–Turner monochromator: light enters through a slit, a mirror makes it parallel, the grating disperses it, a second mirror focuses the spectrum on the exit slit, and turning the grating (mλ = 2d cos φ sin θ) chooses the wavelength. The colours drawn beside the selected one are spread out much more than in a real instrument so you can see them miss the exit slit. The bandpass is slit width × reciprocal linear dispersion; with equal slits the slit function is a triangle of that half-height width. Light through the monochromator grows with the square of the slit width, so narrowing the slits sharpens bands but costs light and adds noise. The holmium oxide spectrum below shows real certified band positions (USP 〈857〉, NIST SRM 2034); the band widths and heights are illustrative. Assumed instrument: 1200 grooves/mm, f = 250 mm, φ = 15°.
Two instrumental effects bend the calibration line at high absorbance. Stray light, a fraction s of the beam that reaches the detector without passing through the sample, gives A = −log[(T + s)/(1 + s)]: with 0.1% stray light a true absorbance of 2 reads 1.959, and readings level off near −log s. A finite bandpass mixes wavelengths with different molar absorptivities; the edges of the slit see weaker absorption and let relatively more light through. When the bandpass is about a tenth of the band width and stray light is 0.1%, the curve stays nearly straight up to A ≈ 2 (O’Haver). The band here is a model Gaussian 20 nm wide centred at 500 nm.
A single-beam instrument reads the blank once and then the sample; if the lamp drifts in between, the absorbance drifts too. A double-beam instrument splits or chops the beam and compares sample and reference continuously, so slow lamp changes cancel in the ratio. The drift shown (−2% per hour plus a slow wobble) is an illustrative model. Counting photons has an uncertainty of √N, so the signal-to-noise ratio grows as √N: ten times more light gives about three times less noise. A photomultiplier measures one wavelength at a time and must scan; a photodiode array records the whole spectrum at once.
Ten samples: methylene blue, chlorophyll b, β-carotene, potassium permanganate, caffeine, retinol, adenosine, anthracene, naphthalene and benzene. Scan from 200 to 800 nm to find λmax, or measure at a single wavelength and compare your sample with five standards. The wavelength of maximum absorbance and the molar absorptivity there are published values (OMLC/PhotochemCAD, the Merck Index, and teaching tables); the band widths and the second bands are illustrative, and real spectra of the aromatic compounds have more fine structure. The reading includes the instrument's bandpass and stray light, so you can see their effect on a real measurement. The colour of the solution is worked out from the light it transmits.
Nine kinds of question rotate: the grating equation, bandpass, overlapping orders, where a lamp peaks, slits and light, stray light, instrument design, signal and noise, and Beer’s law with the sample data. Every number you need is in the Conditions box. A wrong answer is crossed out and you can try again; ★ counts questions right on the first try. Hint gives the method; Show the solution asks twice whether you are sure.
Sellmeier coefficients: Malitson, J. Opt. Soc. Am. 55, 1205 (1965); Schott N-BK7 datasheet. Sample data: S. Prahl, OMLC (PhotochemCAD); Merck Index; King Saud University and VŠCHT teaching tables; Belay et al. (caffeine). Skoog, Holler and Crouch, Principles of Instrumental Analysis (sources, monochromators, stray light, instrumental deviations from Beer’s law); worked grating example: Tel Aviv University physical chemistry laboratory notes; Czerny–Turner relations and the slit-width and Beer’s-law simulations: T. O’Haver, University of Maryland; holmium oxide band positions: USP 〈857〉 and NIST SRM 2034; deuterium lines: European Pharmacopoeia 2.2.25.