ChromFAQ:InternalStandard
From ChromFAQ
What is an Internal Standard?
The basic assumption underlying most quantitative LC is that the area under a peak is proportional to the amount of analyte injected onto the column. In most cases, "proportional" can be understood to mean "linear function of", so that if the peak area for "sample A" is twice the peak area for "sample B", we can conclude that sample A contained twice as much analyte as sample B (if response is non-linear, an appropriate fitting function can often be applied, so that the following discussion applies). If we know how much analyte was present in sample B (the "standard", then we can easily calculate the amount in sample A.
In principle, this procedure can be carried out by generating a calibration plot similar to the one shown here. A series of "calibrators" (often called simply "standards") containing known amounts of the analyte are run and the resulting peak areas (on the y-axis) are plotted as a function of the amount of analyte (on the x-axis). The best straight line is then drawn through the data points. When a sample is run, the area under the analyte peak is measured, and the corresponding amount of analyte is simply read from the calibration plot.
In practice, the entire procedure is carried out computationally. After a set of calibrators has been run, the slope and intercept of the relationship between peak area and analyte amount are computed using a least-squares fit. Assuming the intercept is essentially zero, the slope is called the "response factor" for the analyte. When a sample is run, the area under the analyte peak is divided by the response factor to calculate the amount of analyte. The entire procedure is referred to as external standard calibration, because the calibrators are run separately from the samples. If the word "calibration" is used without other qualification, then external standard calibration is assumed.
In order for external standard calibration to be valid, the calibrators must be run under identical conditions to the samples. This means that a calibration plot is unique to a given system, column, batch of mobile phase, day, etc. It is not necessary to know the exact values for injection volume or detector sensitivity. If chromatographic conditions change however, the calibration must be rerun.
Generally, HPLC systems are consistent enough that external standardization works quite well. In some cases, however, variations in injection volume or detector response from sample to sample are sufficient to compromise the accuracy and precision of the results. In such cases, internal standardization can improve matters.
In internal standardization, an additional compound (called, appropriately enough, the "internal standard") is added to all samples and calibrators. Although different concentrations of the internal standard can be used, in practice is it usually more convenient to use the same concentration in each sample or calibrator. When the calibrators are run, the areas of both the analyte and internal standard peaks are measured, and the ratio of the peak areas is computed. From here on, internal standardization is computationally similar to external standardization. The only difference is that peak area ratios are used instead of peak areas and amount ratios are used instead of amounts. The slope of the calibration plot is called the "calibration factor" rather than the response factor:
CF = (Area ratio)/(Amount ratio)
CF = (AreaA/AreaIS)/(AmountA/AmountIS)
AmountA = (AreaA/AreaIS)*AmountIS/CF
An internal standard is not a panacea. Its effectiveness depends on the source(s) of error in the measurements:
● if the errors in the analtye and internal standard peaks are correlated they will cancel and the precision of the area ratios will be better than the precision of the individual areas. This is most often the case for errors in sample processing, dilution, or injection.
● if the errors in the analyte and internal standard peaks are uncorrelated (independent), they will accumulate and the precision of the area ratios will be worse than the precision of the individual areas. This is most often the case for errors due to chromatographic problems, integration settings, baseline noise, etc.
In principle, any compound can be used as an internal standard, so long as it meets certain critical requirements:
● the internal standard must never occur naturally in the samples or calibrators
● the internal standard must be well-resolved from any other peaks in the samples or calibrators
● if the internal standard is added before or partway through sample preparation, it must behave in a similar fashion to the analyte (i.e., similar recovery, distribution coefficient, solubility, etc.)
● ideally, the internal standard will elute close to the analyte peaks so that any problems are likely to affect both peaks in a similar fashion.
