The Matrix: Everything Else in the Sample
Every real sample is the analyte plus a crowd of other stuff: salts, proteins, minerals, colour. That crowd is the matrix — everything in the sample that isn't the thing you're measuring. In clean lab standards, the matrix is just pure solvent. But seawater, blood, or soil extract carry a thick, complicated matrix, and the instrument doesn't always ignore it.
When the matrix changes how much signal a given amount of analyte produces, you have a matrix effect. Maybe the salts in seawater suppress the analyte's flame emission, so the same iron gives a weaker signal in seawater than in pure water. This is the silent killer of the external standard method: your standards live in clean water, your sample lives in seawater, and the curve you built simply doesn't apply to the sample.
First Defence: Match the Matrix
If the trouble is that your standards live in a different world from your sample, one cure is to give them the same world. Matrix matching means deliberately making your standards in a background that mimics the sample's — adding the same salts, the same acid, the same bulk composition. If both your standards and your sample sit in artificial seawater, the matrix effect hits them both equally and largely cancels out.
Matrix matching works wonderfully when you know what the matrix is and can recreate it. But many real matrices — a particular patient's blood, one specific river — are complicated, variable, and never perfectly reproducible in a clean flask. When you can't faithfully copy the matrix, you need a method that doesn't require you to.
Calibrating Inside the Sample: Standard Addition
Here is the elegant trick. If you can't bring the matrix to the standards, bring the standards to the matrix. In standard addition, you take several portions of the actual sample and spike them with known, increasing amounts of your analyte — adding a little more, then a little more. Every portion already carries the real matrix, so every measurement happens in exactly the world your sample lives in.
Plot the signal against the amount you added. Because the sample already contained some analyte before you added any, the line doesn't start at zero — it starts already lifted off the ground. The amount that was there to begin with is read where the line, extended backwards, would cross zero signal. The matrix effect is baked equally into every point, so it scales the whole line without breaking the answer.
- Split your sample into several equal portions.
- Spike each portion with a different known amount of analyte — for example 0, 1, 2, and 3 added units.
- Measure the signal of each portion and plot signal against the amount added.
- Extend the line back to where signal would be zero; the distance from there to the origin reveals the original concentration.
Choosing Your Approach, and Keeping It Fresh
You now have a small toolkit. Clean samples that behave like clean standards: an ordinary external curve. Wobbling injection volumes or losses during prep: an internal standard. A known but troublesome matrix: matrix-matched standards. An unknown, irreproducible matrix: standard addition. The art is matching the method to how badly the sample misbehaves — and not paying for complexity you don't need.
Whatever you choose, a calibration is not eternal. Lamps dim, columns age, room temperature drifts — and slowly the curve you built this morning stops describing the instrument this afternoon. So you periodically rebuild it, a practice called recalibration. A common discipline is to re-run a known standard partway through a batch: if it still reads correctly, the curve still holds; if it has drifted, you recalibrate before trusting another result.