More Than You Ever Wanted to Know About Calibrations, Part 1 – Types of Calibrations

In previous blog posts on TO-15A and OTM-45 I’ve expressed strong opinions about calibration acceptance criteria. While the acceptance criteria is important, I thought I’d step back and delve more deeply into calibrations, starting with the different types of calibration. A calibration is just a way to convert your instrument response to a known of analyte, and then to a concentration based on injection amount and sample preparation steps, but there are several different flavors of calibration to choose from, and choosing the best one requires you to weigh the advantages and disadvantages. The following post will highlight some of those differences, therefore hopefully enabling you to make the best educated decision for your lab.

Area Percent

The most basic calibration type is an area percent calculation. In fact, it may not even need to be calibrated in the traditional sense, with no response factors needed in some cases. In its simplest form, you simply sum the total areas of all detected peaks and divide the individual peak areas by the total. While it’s the simplest way to get quantitative data, it has several limitations. Since it takes the sum total of the detected peaks as the sum total of the sample, it requires everything in the sample to be detectable by the instrument. This requires the samples to be well understood, and that the sample be suitable for the instrument and detector used. If response factors aren’t generated then all compounds need to have identical responses, or one compound needs to be significantly higher than the rest. Why would response factors be less important when one compound dominates?

Peak	Area	Area %	RF	Adjusted area	Adjusted Area %
A	99.0	99.0%	1.0	99.0	98.9%
B	0.7	0.7%	1.2	0.8	0.8%
C	0.3	0.3%	0.8	0.2	0.2%
Peak	Area	Area %	RF	Adjusted area	Adjusted Area %
A	33.3	33.3%	1.0	33.3	33.3%
B	33.3	33.3%	1.2	40.0	40.0%
C	33.3	33.3%	0.8	26.6	26.7%

Table 1 – Comparison of Area % results with and without response factors

If we look at Table 1 we see two examples, one with a mix of 3 peaks with Peak A much greater than Peaks B and C, and one with all peaks having equal responses. If a known amount of each compound is injected then a response factor can be generated for each analyte. When the areas are adjusted by this response factor you can see that the adjusted area % closely matches the standard area % for the first scenario, since the relative amounts that Peaks B and C change compared to Peak A is not very significant. However, in the second scenario the results change significantly, since the relative changes are more pronounced when the area counts are similar. This means that for measurement of highly pure compounds response factors may not be needed.

External Standard

The next simplest calibration is the external standard calibration. In this type of calibration a response factor is generated that directly relates the instrument response to the concentration of the analyte. This means the calibration does not require the sample to sum up to 100%, so it can be used when not every component of the sample can be detected by the instrument. This makes it more flexible in the types of samples that can be analyzed and the types of instruments and detectors used. However, instrument drift can change response factors over time, and differences in sample matrix can potentially suppress or enhance response factor as well.

Internal Standard

Internal standard calibrations add a known amount of non-target compounds to the sample post-extraction but prior to analysis. The relative response factor (RRF) generated is then relates the concentration (C_S) and response (A_S) of the target compound to the internal standard response (A_IS) and concentration of the internal standard (C_IS), rather than directly relating the analyte concentration and area. The equation for this is given as RRF = (A_S x C_IS) / (A_IS x C_S).

The addition of internal standards corrects for response changes due to instrument drift and matrix effects. However, for this to work well the internal standards must behave in a similar matter to the target analytes. If the internal standards behave differently than the target analytes then the response corrections may be incorrect, so proper choice of internal standards is critical for this type of method. Since internal standards should closely match the behavior of target compounds, they often will coelute with them, especially if isotopes are used. For this reason, internal standard methods are more common on mass spec methods than with other detectors.

Extracted Internal Standard / Isotope Dilution

This calibration type is similar to the internal standard calibration listed above, but the standards are added pre-extraction rather than post extraction. This means that the calibration can potentially correct for extraction efficiencies. For example, if the internal standard recovers at 50% then the compounds are adjusted by a factor of 2 to correct for the poor extraction efficiency.

If you’re familiar with the use of surrogates for monitoring extraction efficiency in external or internal standard methods, this may seem familiar. A surrogate is a compound not found in the sample (often an isotope or a fluorinated or brominated analogue) that is spiked pre-extraction. The recovery of the surrogate can give you an indication of any loss during extraction, though it doesn’t correct the analyte results. For example, if your result for your target compound is 5 ppb and your surrogate recovers at 70%, you know that your results are biased low. However, if that surrogate compound is instead an extracted internal standard, the results will be adjusted by the 30% loss, so your result will be 6.5 ppb. Post-extraction internal standards are often added as well to monitor the recovery of the extracted internal standards, so they pull double duty as internal standards and surrogates.

Similar to the internal standard method, it’s very important that the extracted internal standards behave similar to the targets in the extraction process, so isotopically labeled versions of the target analytes are frequently used, which is why it’s often referred to as isotope dilution. If the standards extract differently than the target analytes then error can be introduced in the final results due to the improper correction. Matching every target with an appropriate isotope analog can get expensive however, and some compounds may not have isotopes commercially available. Similar to the post-extraction internal standard methods, pre-extraction internal standard methods are generally more applicable to mass spec detectors, and isotope dilution methods require them.

Known Standard Addition

This calibration type is one I’ve seen more commonly with techniques such as titrations rather than chromatography, but I believe it has potential use in GC and LC as well. To do this, an unknown sample is analyzed and a series of known spike amounts of increasing concentration are added. If the spike amount added vs. area is plotted, the intercept of the graph will be the negative of the initial value of the sample, as shown in Figure 1.

Figure 1 – Known standard addition example

Here, the X-intercept would be at -12.5 ppm, giving the initial value of the sample to be 12.5 ppm. The downside of this method is that multiple spikes of a single sample isn’t terribly efficient for chromatographic techniques, though this could be used to generate a quick matrix matched calibration if you don’t have a blank matrix that is truly free of the targets of interest.

While it might seem that extracted internal standards would always be the best choice since they can correct for instrument, matrix, and extraction issues, there’s always the balance of effort vs. data quality. For samples that extract with high efficiency and show little response drift, a simple external standard calibration could provide the same data quality at less cost and work than internal standard or isotope dilution methods. For the analysis of highly pure products an uncalibrated area % method may be able to give adequate results with very little effort. Written methods will usually specify the appropriate calibration type, but if you’re building your own method choosing the right calibration requires you to know your samples, determine the level of data quality needed, and balance that against the resources needed for each calibration type.

Stay tuned for the next blog post, where I’ll discuss curve fit types and weighting.

View all of the posts in the "More Than You Ever Wanted to Know About Calibrations" series.