A spectrophotometer is one of the first instruments a new laboratory worker is trusted with alone, and one of the last they properly understand. It is easy to operate: put a blank in, press a button, put the sample in, read a number. It is also easy to get a plausible-looking number that is completely wrong, and the instrument gives no indication when this happens.
The reason is that the measurement is indirect. The instrument does not detect the substance being quantified. It compares how much light of a chosen wavelength arrives at a detector with the sample in place against how much arrived without it, and reports the loss. Turning that loss into a concentration requires a chain of assumptions, and when any link in that chain fails, the display carries on showing digits with the same confidence as before.
Key takeaways
- Absorbance is a logarithmic measure of light lost, not a direct measure of anything in the sample.
- The Beer-Lambert relationship holds only when molecules absorb independently and the light is genuinely monochromatic.
- Readings become unreliable at both extremes: too little absorbance is dominated by noise, too much by stray light.
- Turbidity scatters light and inflates absorbance across all wavelengths, which is the most common cause of a silently wrong result.
- Nucleic acid purity ratios detect classes of contaminant; they say nothing about whether the material is intact.
Light Absorption as a Concentration Proxy
Molecules absorb light when a photon’s energy matches the gap between two of their electronic energy levels. The electron is promoted to a higher state, the photon disappears, and its energy is eventually dissipated as heat or re-emitted. Because the available energy gaps are a property of molecular structure, each compound absorbs a characteristic set of wavelengths, and that pattern is its absorption spectrum.
What the instrument measures is transmittance: the fraction of incident light that makes it through to the detector. Transmittance relates to concentration in an awkward way, because absorption is multiplicative rather than additive. Each successive thin layer of solution removes a fixed proportion of the light reaching it, not a fixed quantity, so transmittance falls exponentially as concentration or path length increases.
Absorbance is defined as the negative logarithm of transmittance precisely to undo that exponential. Taking the logarithm converts the multiplicative relationship into an additive one and produces a quantity that rises in proportion to concentration. This is why absorbance and not transmittance is the working unit, and it also explains the scale: an absorbance of one means a tenth of the light got through, two means a hundredth, three means a thousandth. Each additional unit of absorbance corresponds to a tenfold reduction in the light reaching the detector, which will matter shortly.
The Beer-Lambert Relationship
The working equation states that absorbance equals the product of three things: a constant characteristic of the substance at that wavelength, the length of the light path through the solution, and the concentration.
The substance-specific constant, the molar absorption coefficient, expresses how strongly one mole per litre of the compound absorbs over a one centimetre path. It is a fixed property of a given molecule at a given wavelength in a given solvent, and its magnitude varies over an enormous range. Highly conjugated dyes and certain protein cofactors absorb intensely, which is why they can be measured at very low concentrations. Compounds with weak absorption require correspondingly higher concentrations to produce a usable reading, and this alone determines whether direct spectrophotometry is a viable approach for a given analyte.
Path length is usually one centimetre by convention, which is why standard cuvettes have that internal width and why published coefficients assume it. Concentration is what is being sought.
The relationship’s convenience conceals three assumptions that are worth stating explicitly, because every failure mode later in this piece is one of them breaking.
The first is that each absorbing molecule behaves independently of every other. The second is that the light passing through is monochromatic, meaning a single wavelength rather than a narrow band. The third is that the only thing happening to the light is absorption by the analyte, with nothing scattering it, nothing else absorbing at that wavelength, and no light reaching the detector by any route other than through the sample.
Choosing the Right Wavelength
The conventional advice is to measure at the wavelength of maximum absorbance, and the reasons are worth understanding rather than accepting.
Sensitivity is the obvious one: the peak gives the largest signal for a given concentration, so the smallest concentration difference is detectable there. The subtler and more important reason concerns wavelength accuracy. At the top of a peak the spectrum is flat, so a small error in the wavelength the monochromator actually delivers produces a negligible change in measured absorbance. On a steep flank, the same wavelength error produces a substantial change. Measuring at a peak makes the result robust against the drift and calibration error that every instrument has.
Bandwidth is the associated instrument parameter and is frequently overlooked. No monochromator delivers a single wavelength; it delivers a band whose width is set by the slit. If that band is narrow relative to the width of the absorption peak, the Beer-Lambert assumption of monochromatic light holds well. If the band is broad, the instrument is averaging absorbance across a range where it varies, and because absorbance is logarithmic, that average is systematically low. Sharp peaks with a broad slit therefore produce readings that are both depressed and non-linear, and the effect worsens as concentration rises.
Scanning a full spectrum before quantifying is cheap and repays the effort. It confirms the peak is where it should be, which is a check on both the compound’s identity and the instrument’s wavelength calibration. It reveals shoulders that suggest a second absorbing species. And it shows whether absorbance rises towards the short-wavelength end without a peak, the signature of scattering rather than absorption.
Blanking and What It Corrects For
The blank defines zero, and everything downstream inherits whatever the blank got wrong.
A blank is a sample identical to the one being measured in every respect except that it lacks the analyte. Measuring it establishes the reference intensity against which the sample is compared, and it therefore subtracts the absorbance of the solvent, of buffer components, of any reagent added during a colour-forming reaction, and of the cuvette walls themselves.
The requirement that the blank match the sample in everything but the analyte is stricter than it sounds. A blank made with water when the sample is in buffer will not correct for buffer absorbance, which in the far ultraviolet can be considerable. A blank that omits a reagent present in the sample will not correct for that reagent’s own colour. A blank measured in a different cuvette from the sample will not correct correctly, because two cuvettes are never optically identical, and a blank measured in a cuvette rotated differently will not either, since cuvette faces vary slightly around their circumference.
There is a further point about what blanking cannot do. Blanking corrects for absorbance that is present equally in blank and sample. It cannot correct for anything that appeared only in the sample, which includes turbidity from precipitated material, contamination from the extraction, and any coloured breakdown product. These are exactly the things that go wrong, and they are exactly the things blanking is powerless against.
Where Linearity Breaks Down
Absorbance is proportional to concentration over a limited range, and both ends of that range have distinct causes of failure.
At low absorbance the problem is signal to noise. The instrument is comparing two large light intensities that differ only slightly, and the difference eventually becomes comparable to the electronic and photon noise in the measurement. The relative error therefore grows rapidly as absorbance falls, which is why a reading close to zero is not a precise statement that very little is present; it is an imprecise statement.
At high absorbance the dominant problem is stray light. Some small proportion of the light reaching the detector has never passed through the sample properly, having scattered off internal surfaces or leaked past the monochromator. When the sample transmits plenty of light, this stray component is negligible. When absorbance reaches two or three, the genuine transmitted light has fallen by a factor of a hundred or a thousand while the stray light has not fallen at all, so it becomes a substantial share of what the detector sees. Measured absorbance flattens out, approaching a ceiling determined by the instrument’s stray light level, and the curve bends towards the horizontal.
| Region | Typical absorbance | Dominant limitation | What the reading does |
|---|---|---|---|
| Very low | Below about 0.1 | Detector and photon noise | Imprecise, poor relative accuracy |
| Working range | Roughly 0.1 to 1.0 | None dominant | Linear and reliable |
| Upper margin | Roughly 1.0 to 2.0 | Stray light, bandwidth effects | Slight negative curvature |
| Saturated | Above about 2.0 | Stray light dominates | Flattens towards a ceiling |
Scattering is the failure that catches most people, because it does not look like a failure. Particles suspended in the solution scatter light away from the detector, which the instrument records as absorbance. The characteristic signature is elevated readings across the whole spectrum, rising steadily towards shorter wavelengths with no peak. A simple check is to read at a wavelength where nothing in the sample should absorb at all; a clean solution reads near zero there, and anything appreciably above zero indicates turbidity contaminating every other reading on the sample.
The practical remedy for all of these is the same and is unglamorous: dilute the sample into the working range, measure, and multiply. A dilution is a small amount of extra work and it converts an unreliable reading into a reliable one.
Cuvette Materials and Path Length
The container is an optical component and behaves like one.
Material determines the accessible wavelength range. Ordinary glass and most plastics absorb strongly in the ultraviolet, becoming effectively opaque below roughly the middle of the ultraviolet region, so they are usable for visible work and useless for nucleic acids and proteins measured near their ultraviolet peaks. Quartz transmits far into the ultraviolet and is the required material for that work. Specialised low-ultraviolet plastic cuvettes extend the plastic range partway and are convenient disposables where cross-contamination is a concern, but their transmission should be verified rather than assumed.
Path length scales the signal directly. Doubling the path doubles the absorbance for the same concentration, which offers a straightforward route out of the noise-limited region for dilute samples. Long-path cells are used for exactly this. Short-path cells do the opposite and are the basis of the microvolume instruments now common in molecular biology, which hold a droplet between two pedestals and use a path of a fraction of a millimetre. These extend the measurable concentration range upward by a large factor without dilution, at the cost of greater sensitivity to droplet formation, evaporation and surface cleanliness.
Handling completes the list. Fingerprints absorb and scatter, so cuvettes are held by the frosted faces. Bubbles adhering to the optical surface scatter strongly and are a common cause of an anomalous single reading that vanishes on repeat. Scratches from abrasive cleaning are permanent. Matched cuvette sets exist because two nominally identical cuvettes differ measurably, and where the highest accuracy is needed, blank and sample are read in the same cuvette in the same orientation.
Nucleic Acid Purity Ratios in Practice
The most common routine application of ultraviolet spectrophotometry is estimating nucleic acid concentration and purity, and it is also the application where the readings are most often over-interpreted.
Nucleic acids absorb maximally near 260 nanometres because of their aromatic bases. Proteins absorb near 280 nanometres because of tryptophan and tyrosine. The ratio of absorbance at these two wavelengths therefore gives a rough indication of protein contamination, with values around 1.8 conventionally regarded as acceptable for DNA and around 2.0 for RNA, RNA reading higher because of its base composition.
A second ratio compares 260 nanometres against 230, where several common laboratory contaminants absorb: guanidinium salts from extraction kits, phenol, and some carbohydrates. A low value here points to carryover from the purification chemistry, which matters because these compounds inhibit the enzymes used in downstream reactions even when they contribute little to the concentration estimate.
Concentration itself is calculated using established conversion factors relating absorbance at 260 nanometres to mass concentration, with different factors for double-stranded DNA, single-stranded DNA and RNA. These are averages over typical base composition and are approximations for any particular sequence.
The limitations deserve as much attention as the numbers. The 260 reading counts every nucleic acid present, so it cannot distinguish the intended DNA from contaminating RNA, from free nucleotides, or from degraded fragments. A thoroughly degraded sample and an intact one of the same mass give the same reading. The ratios are insensitive at the levels that matter for many downstream applications, since a small quantity of a potent inhibitor can ruin a reaction while barely moving a ratio computed from a large nucleic acid signal.
Where the answer must be specific rather than indicative, a fluorescence-based assay using a dye that becomes fluorescent only when bound to the target species is the appropriate tool, since it responds to double-stranded DNA in the presence of RNA and free nucleotides. Where integrity matters, an electrophoretic or capillary sizing method is required, since no absorbance measurement can report fragment length.
The general discipline that prevents most spectrophotometry errors is short. Scan the spectrum rather than reading a single point, and look at its shape. Keep readings inside the linear working range by diluting rather than trusting a high number. Blank with the exact matrix the sample sits in. Check a wavelength where nothing should absorb, as a turbidity test. And remember what the instrument is actually reporting: light that failed to arrive, for whatever reason, of which the concentration of interest is only one.
Frequently asked questions
Why should absorbance readings be kept below about one?
Because above that level a small and constant amount of stray light inside the instrument becomes a large fraction of the light reaching the detector. Genuine transmitted light falls tenfold for every unit of absorbance, while the stray component does not fall at all, so the measurement progressively understates the true value and the response curve bends towards a ceiling. Bandwidth effects and molecular interactions add further curvature in the same direction. Diluting the sample into the range where the relationship is linear costs a few minutes and removes all of these problems at once.
My purity ratio is fine but my downstream reaction failed. Why?
Purity ratios are insensitive to small quantities of potent inhibitors. A residual trace of a chaotropic salt or an organic solvent can inhibit a polymerase while contributing almost nothing to absorbance at 230 nanometres relative to a strong nucleic acid signal at 260. Ratios also say nothing about integrity, so a heavily degraded preparation can have textbook ratios. Additional cleanup, or a fluorescence-based quantification alongside a sizing method, generally resolves the discrepancy faster than repeating the absorbance measurement.
What does it mean when absorbance rises steadily towards shorter wavelengths with no peak?
That pattern is the signature of light scattering rather than absorption, and it means the sample contains suspended particles: precipitate, aggregated protein, cell debris or air bubbles. Scattering removes light from the beam at every wavelength and therefore inflates every reading on that sample, including any ratio calculated from them. Centrifuging or filtering the sample and remeasuring is the standard response, and reading at a wavelength where nothing should absorb makes a quick diagnostic check before quantifying anything.
Can I use a plastic cuvette for nucleic acid measurements?
Standard polystyrene and most other common plastics absorb strongly in the ultraviolet and are unsuitable near 260 nanometres, where they will produce readings that are dominated by the cuvette rather than the sample. Quartz is the traditional answer. Specific ultraviolet-transparent plastic cuvettes exist and work, but their transmission range varies by manufacturer and should be checked rather than assumed. A simple test is to blank against air and read an empty cuvette across the wavelength range of interest; a material that is opaque there makes the problem obvious immediately.
Is a microvolume instrument as accurate as a conventional cuvette?
For samples comfortably within its range, it is generally comparable and considerably more convenient, since it needs only a droplet and no dilution for concentrated samples. Its very short light path extends the upper concentration range substantially, but the same short path makes it less sensitive at low concentrations, where a conventional or long-path cuvette performs better. It is also more sensitive to technique: the droplet must form a proper column between the pedestals, the surfaces must be genuinely clean, and evaporation during a slow measurement concentrates the sample. Replicate readings on the same sample reveal most of these problems quickly.
