What is actually calibrated?
In external calibration, solutions with known analyte concentrations establish the relationship between response and concentration. The sample is then interpreted using that relationship. This requires sufficiently comparable preparation, injection and detector behavior between standards and samples. External calibration can be appropriate when those conditions are controlled and the intended performance is demonstrated.
For internal calibration, a distinguishable standard is added to calibrators, controls and samples. The response ratio reduces variability that affects both substances proportionally. UV methods need signals that can be measured separately; LC–MS can sometimes distinguish coeluting isotope-labelled compounds by mass. The choice is analytical, not a rule that internal calibration is always superior.
Agilent OpenLab: internal-standard calculation and response ratios · ICH M10, §3.1–3.2: reference standards, matrix effects and calibration (FDA)
Follow one calculation through to the original sample
Synthetic example with a constant internal-standard concentration: let R = Aₓ/Aᵢₛ and fit R = 0.020 × c + 0.010, where c is the analyte concentration in µg/mL in the prepared measurement solution. For R = 0.410, c = (0.410 − 0.010)/0.020 = 20.0 µg/mL. The slope therefore carries units of mL/µg; the response ratio is dimensionless.
If the original solution was diluted tenfold to make that measurement solution, its concentration is 200 µg/mL, assuming the dilution is the only additional conversion. This does not establish the amount in a solid sample or vial: those require the original weighing, extraction volume and the procedure's defined calculation. If internal-standard concentration varies, use the calibrated amount-ratio model instead of silently treating it as constant.
Synthetic calculation with a fixed internal-standard level
On narrow screens, scroll the table sideways. You can focus the table and use the arrow keys.
| Stage | Value | Meaning |
|---|---|---|
| Measured response ratio | 0.410 | Analyte area divided by internal-standard area |
| Calibration result | 20.0 µg/mL | Prepared measurement solution |
| Tenfold dilution correction | 200 µg/mL | Original solution under the stated assumption |
What the ratio corrects—and what it cannot
Suppose a smaller injection reduces both analyte and standard areas by 10%. Their ratio is unchanged. If an unresolved interferent instead increases only the analyte area, the ratio is biased. If analyte is lost during extraction but the standard is added afterward, the standard never experienced that loss. The timing of addition therefore determines which steps it can monitor.
Review the raw internal-standard response as well as the ratio. An apparently normal ratio can conceal both signals becoming very small, while an unusually high standard response may indicate an addition error or interference. Suitability includes stability, absence of unwanted contributions in the measured channel and behavior across the intended matrix and range.
Waters: Stable-Isotope-Labeled Internal Standards and Matrix Effects · Agilent OpenLab: internal-standard calculation and response ratios
Check the method beyond the fitted line
A calibration model needs an appropriate response relationship and working range. A high correlation coefficient alone does not show that concentrations are correct throughout that range. Examine residual patterns and back-calculated standards, and use suitable independent control preparations to test the procedure. Do not force a zero intercept, weighting scheme or wider range solely to make the plot look better.
Keep stock-standard identity and assigned concentration, preparation records, units, blank treatment, internal-standard addition point and sample dilution factors with the method. Trace a result from the detector signal through each conversion. For an out-of-range sample, a justified dilution and reanalysis may be appropriate; extrapolation is not established merely by extending the plotted line.
ICH M10's recommendations concern bioanalytical study samples, while other intended uses need their own justified design. Calibration connects signal and amount under stated conditions; validation and ongoing run checks establish how reliably the full procedure performs that task.