Errors in Method Transfer What if Two Chromatograms do not Look Perfectly Alike?
High-performance liquid chromatography (HPLC) methods are commonly developed on instruments different to those used for routine application. The following article describes the challenges this poses for users.
HPLC systems can vary in specification and configuration, including the formation of elution gradients or dwell volume, mixing of sample zone with mobile phase, extra band dispersion, mobile phase pre-heating, column thermostating, parameters of the detector, and data processing in the detector firmware. Therefore, method transfer between instruments may result in differences in the chromatograms, even when the column and method-related parameter settings are kept the same. Even though superimposable chromatograms are not typically a requirement, it is essential that a method passes the system suitability test (SST) when employed on a new HPLC system. The SST demands sufficient peak resolution, signal-to-noise performance and detection linearity with all relevant analyte peaks of the chromatogram. As a result, an HPLC method should only be made available with proven robustness against all variations that different instruments can possibly introduce. In practice, however, this is more a theoretical ideal than a practical reality.
Root cause analysis of SST failure
It is important to understand why a chromatogram on an instrument deviates from the original for a given method. When the root cause is identified, a successful solution can be implemented and proven by passing the SST, ensuring the system can be successfully deployed for the desired method.
A meaningful SST report should provide sufficient diagnostic information to reveal whether a failure was due to insufficient peak resolution or another root cause. If minimum peak resolution was not achieved, it must be verified whether retention times have shifted resulting in merged peaks, or whether peak shapes relating to their widths and symmetry are different. Once this is diagnosed, the flow schemes in Figure 1 help to identify the cause of this symptom. Assuming a change in relative retention time has been identified, the scheme in Figure 1A can be followed for further diagnosis.
Distinguishing between isocratic and gradient methods is a key first step. If an isocratic method is run on a gradient instrument and the mobile phase composition is accomplished by programming the eluent composition, a proper proportioning (by pre-mixing of the mobile and pumping with only one channel) must be verified. If this does not show any difference, the root cause for the deviating relative retention likely stems from temperature effects. This can be confirmed by assessing the impact of the column oven temperature on the relative retention. If the method in question uses a gradient, both the gradient formation or gradient delay and the internal column operating temperature could be likely suspects.
To distinguish gradient effects from column thermostating issues, the method should be transformed to an isocratic test mode. To do this, the elution composition of the first analyte peak pair that shows inferior resolution should be determined and prepared as the isocratic mobile phase. The mobile phase can then be applied to the isocratic elution of the SST sample on both the originator and the target instrument. If the results generated are now identical, the root cause was the gradient formation. If the results still differ, then the column temperature influence should be investigated further.
However, if a difference in the gradient formation behavior occurs, the effective gradient dwell volume/gradient delay volume (GDV) has to be determined on both systems. A GDV deviation can change relative retention and elution order in certain applications (1). Standardized GDV tests like the classical Dolan test (2) also provide insights into the rounding of edges in the gradient formation that result from gradient mixers in the pump. Once a system has been successfully modified to match GDV, or a match was achieved by shifting the sample injection time relative to the gradient start, the method transfer issue should be resolved.
In the rare cases when the problem persists, the reason may be a difference in the gradient formation. Mixing of the mobile phase components can occur either on the low-pressure side in front of the pump head (LPG pumps) or on the high pressure side behind the individual pump heads (HPG pumps) and this may cause differences in flow and composition. In other words, transfers from HPG to LPG or vice versa should be avoided whenever possible.
If the (relative) retention times match perfectly between the two systems, but peak resolution does not meet experimental requirements, a poorer peak shape may be the reason. A multitude of instrument parameters can affect peak shape, but the most likely are column thermostating, mobile phase pre-heating, or the fluidics behind the injector. The scheme presented in Figure 1B outlines how to identify more complex root causes.
As a first step, the detector setting influence should be investigated, as a higher filter constant of the electronic filter can cause broader and more tailing peaks. If reduction of the filter constant does not help to increase the resolution, a smaller injection volume (reduction by factor 2 or more) should be tested. If this improves the resolution, the poor mixing of the sample zone in front of the column is the most probable root cause. This is only a plausible explanation if the sample is dissolved in a solvent of stronger elution power than the mobile phase. If such a solvent mismatch effect can be excluded, the root cause is likely caused by extra column dispersion or the radial temperature gradients inside the column. These effects can be easily distinguished when using an isocratic method by plotting the plate number N of all peaks recorded in the SST against their retention factors k. A more pronounced increase of N with increasing k points to an increased extra column dispersion, while a decrease of N, in particular at higher retention (above k = 2) indicates thermal mismatch effects.
If the signal-to-noise ratio is not sufficient on the new system, Figure 1C provides a suitable evaluation approach, starting with the distinction between increased noise versus decreased peak height. Knowing which phenomenon applies, inspection of the different root causes affecting peak height and noise is required.