Selection of the optimal column is often decisive in the success or failure of method development. As the first step in developing an analysis protocol, the analyst must select the best RPLC column for the specific application. There are several considerations in column selection, each dictated by the demands of the separation problem. An introduction to column selection and the effect of column parameters is explained in the Topic "Columns".

Obviously, candidate columns are selected to meet these demands to the greatest possible extent. In analytical practice, chromatographers need tools to aid in the selection of the proper column for the given analytical problem. Apart from financial considerations, factors influencing column selection will include:

• Retention and selectivity
• Efficiency and peak symmetry
• The average pore size and its distribution over the stationary phase
• Sample loadability
• Chemical and thermal stability
• Batch to batch reproducibility and long term availability

Knowledge of the chemistry of column preparation is particularly important since stationary phase stability plays more and more of a role in column selection as analysts push the limits of RPLC. These aspects are described in other chapters within this Topic Circle.

Ideally speaking, column selection should be based on objective criteria, of which the retention and selectivity would be the most important parameters.

Column selection should be based on objective criteria, the most important parameters are:  

1 Retention and selectivity
Retention and selectivity are of paramount importance in selecting RPLC columns. Consider that retention combines with efficiency to determine the resolution, in addition do determining analysis time (and therefore sample throughput).

2 Efficiency and peak symmetry
The net efficiency of a column, i.e. the number of theoretical plates, is determined by the stationary phase particle size, column length, and also by the symmetry of the chromatographic peaks.  Useful information on the properties of a column, including particle size, column dimensions, and efficiency can be found in the column manual.

3 The average pore size and its distribution over the stationary phase
The average pore size determines the extent of analytes’ access to the pore volume, and thus, to the active chromatographic area of a stationary phase. In general when the pore size of the packing becomes less than ten times the molecular size of the analyte, the resultant restricted diffusion may decrease column performance significantly since not all of the chromatographically active sites are accessible. Once again, the column documentation provides useful information on such bulk properties.

4 Sample loadability
To a first approximation, the amount and volume of sample that can be loaded on a column depends on the available chromatographic surface area. The sample loadability is also influenced by the nature and the polarity/ionic state of the analytes. Generally, in order to take full advantage of the resolving potential of a column (working under linear adsorption isotherm conditions), sample mass and volume overloading should be avoided. Upon decreasing the available surface area of a column, the sample loadability will also decrease drastically. Other phase types, for example nonporous and pellicular, have much lower specific surface areas than their fully porous counterparts. Therefore, in order to take advantage of their resolution potentials, the volume and mass loading limits of the different types of columns must be respected.

5 Chemical and thermal stability
There are several reasons to use more chemically and thermally stable columns. For example, working at elevated temperatures can reduce analysis times, while using certain more aggressive eluents can improve selectivity. Clearly, therefore, the stability of the column is an important consideration. A more detailed discussion on the chemical and thermal stability of RPLC columns can be found in the paragraph "Chemical and thermal stability".

6 Batch to batch reproducibility and long term availability
Analysts want to make sure that replacement of a column with another of the same brand will not disturb a routine analysis. In other words, column reproducibility is very important. This is especially true when validated analysis protocols are used or when performing interlaboratory studies. In recent years manufacturers have made substantial progress in the manufacture of highly reproducible RPLC phases. This issue is discussed in more detail in the paragraph "Reproducibility and repeatability".

1. Determination of the physical-chemical properties of stationary phases, including average particle and pore size, pore volume, carbon load, ligand density, and mechanical strength.
2. Spectroscopic techniques, such as infrared (IR)-, fluorescence (FL)-, and solid state nuclear magnetic resonance ( NMR ) spectroscopy. These techniques are particularly useful for providing information about ligand attachment and conformation.
3. Thermodynamic measurements, which give information on the enthalpic and entropic contributions to a separation process.
4. Chromatographic tests providing information on parameters such as the hydrophobicity, polar and ionic activity, shape selectivity, and metal activity of stationary phases.

Although the determination of physical -chemical, spectroscopic and thermodynamic properties is of great importance, consensus now exists that only chromatographic tests can reveal the subtle differences between RPLC phases. Because RPLC is used in so many application areas for the analysis of samples of such varying chemical and physical nature, a single chromatographic test usually fails to give a complete picture of the chromatographic properties of columns. Thus, current thinking holds that a number of chromatographic tests are necessary for a complete characterization and ranking. The current tests for characterizing RPLC columns chromatographically fall into two categories: empirically based and model based test methods.

These methods are linked by their use of different arbitrarily selected test probes which are assumed to reflect specific chromatographic properties such as hydrophobicity, silanol activity, shape selectivity, etc. Well known examples of such testing methods include tests developed by Tanaka, Engelhardt, Walters, Galushko, Eyman, Daldrup, and others. The many in-house column test methods also fall into this category. There are still substantial differences between these methods in terms of both measurement conditions and calculation procedures.

A variety of different empirical approaches are used to measure the chromatographic properties of RPLC columns. A summary of several oft-used empirical tests is given in this pdf (Click to download column comparison).

Coefficients of the mutual correlation of the silanophilic activity results of the E, E_m, G and W- tests and the hydrogen bonding capacity results from the T-test are presented in this table:

Y = a₀ + a₁ X; r = correlation coefficient.

With the exception of hydrophobicity measurements, the correlation between other chromatographic test results and the results of these tests is usually poor. As an example this is illustrated in the above comparison where the differences between the ionic and polar properties obtained from a number of tests are tabulated. As can be seen from that figure, hardly any correlation exists between the silanophylic activity and other quantitated parameters of the different tests. In chromatographic practice this shortcoming seriously complicates the adequate selection of rplc columns. In the chapters below this subject is discussed in more detail and suggestion to solve this problem are also given.

Furthermore, many column tests employ small molecules, giving results that are not necessarily relevant to the separation of large molecules. It has become clear to the RPLC community, therefore, that multiple tests must be used to generate an adequate description of the relevant column properties. Amongst others requirements, such test should include test components which are similar or from the same chemical family. The compiled results of these empirical tests are not necessarily easy to interpret and apply to subsequent analytical problems. As a consequence, analysts often find themselves confronted with non -uniform and confusing information about the properties of their RPLC columns. More examples of the empirical tests developed for the characterisation and ranking of RPLC columns are in preparation for Chromedia. Apart from the testing protocols, these articles also contain column ranking information for a substantial number of modern RPLC phases.

These methods are based on specific physical and chemical models. Such methods seek to describe column properties in terms of objective physico -chemical parameters, which should allow for more objective column characterisation and ranking procedures in comparison to empirically based methods. Model based methods include the interaction model of Jandera, the silanol scavenging model developed by Horvath, the solvatic computional model of Galushko, and quantitative structure retention relationship (QSRR) models. This last model has garnered much attention of late. QSRR‘s are a special form of linear free energy relationships (LFER). In such relationships, a free energy-related property, e.g. a retention factor or distribution constant, is described as a linear function of the intermolecular interactions (descriptors) of a solute and the complementary column plus eluent (system) constants. The general form of a QSRR can be described as follows:

Where “AP is a property for a series of analytes in a fixed solvent system (e.g. retention factor), R₂ is an excess molar refraction, p  is the analyte dipolarity / polarizability, α and β are the analyte’s overall or effective hydrogen-bond acidicity and basicity and V is the McGowan characteristic volume.

The coefficients c, r, s, a, b, and v in this equation are characteristics of the systems. They represent the difference in individual properties (complementary to R, π, α, β and V) between the mobile and the stationary phase. Hence, r should be proportional to the difference between excess molar refractivity of the stationary and the mobile phase, s should reflect the corresponding differences in dipolarity / polarizability, and v, the difference between the McGowan volumes. The coefficient b is assumed to be proportional to a difference in hydrogen bond basicity between the stationary and the mobile phase and a is related to analogous difference in hydrogen-bond acidity. “

Text between “ “ is a citation from R. Kaliszan et al in J.Chromatography A855 (1999) 455-486

Recently, QSSR ‘s and, more particularly, the so called Hydrophobic Subtraction ( HS )model have been studied for potential use in testing, characterising, and ranking RPLC columns according to objective physico -chemical parameters. In the HS model, RPLC columns are described by the parameters: hydrophobicity, steric resistance to insertion of bulky solutes into the stationary phase, column hydrogen bond acidity, column hydrogen bond basicity, and column cation exchange activity. In recent years a substantial number of RPLC columns have been tested, characterised and ranked using the HS model.