All You Need is Selectivity
Recent Developments in Silica and Bonding Chemistry for HPLC
During the last decade, significant advances in chromatography kinetics have been widely applied in using sub 2 micron and core shell media. When coupled with ultra-high-pressure systems (> 400 bar) they yield substantially faster analysis than conventional 5 micron media, although the ultimate resolution is more limited. This stems from the fact that the pressure drops required to deliver a theoretical plate is inversely proportional to particle size. Thus, for a fixed pressure, a greater number of plates can be realised from larger particles but at the expense of generally unacceptably long analysis times.
Developments in the thermodynamics over the same period have been more restrained. Although a number of unusual phases have been developed, the standard C18 and C8 phase remain the main choice due to their versatility and familiarity. However, there is a need to provide a greater range of predictable selectivities within the C18/C8 structures.
Most of the commonly available C18/C8 phases are based on high purity silicas and produce what is commonly termed “Type B” phases. Usually, they are bonded to a maximum level of octadecyldimethyl silane and extensively endcapped. These show generally excellent performance on a wide range of compounds, particularly with strong bases which often feature in pharmaceutical samples. The common design features do however tend to result in relatively similar selectivities. This is excellent news if a backup phase is needed for a separation, but less so if an alternative selectivity is required.
The Type B silicas mentioned above virtually supplanted the previous generation of Type A silicas. These are made from commercial silica sols and give product with low but chromatographically significant levels of trace metals. As an aside, it is noted that they still are >99.5% silica and would pass as analytical standard grade for most chemicals.
Advantages of Type A Silicas
Significant advantages of type A silicas have been identified when used at high pH. Kirkland showed that the thicker walls of typical sol gel type A silicas gave greater stability with aggressive high pH eluents than sil-gel type phases and the imbedded trace metals lead to reduced solubility under similar conditions .
The latter is presumably linked to the aluminium content which mimics the high pH stability of aluminosilicates.
Although the demise of Type A media has often been anticipated, there remains a market for such products due to the different selectivity they provide. What was needed was a Type A media which addressed the perceived problems (base tailing, reproducibility) while offering an extended range of selectivity. Both would lead to an increased predictability and would significantly differ from the conventional Type B phases.
After many years of being only available to original equipment customers, Exsil columns which provide access to these media are available to the end user. This type of silicas were initially developed in the mid 1980’s and use a novel manufacturing procedure which allows the accurate and independent control of surface area, pore volume and pore diameter. Typically, the 100 A media is controlled between 97 and 103 A with pore volume between 0.50 and 0.52 cc g-1.
Significant product introductions through this period include the following materials:
- Sub 5um wide pore (1985)
- Sub 2um media (2004)
- Sub 2um wide pore (2004)
The early silica lines produced Type A silicas, while these have been supplemented by high purity type B phases. The attributes of Type A silicas have been overshadowed by the Type B, but they offer significant opportunities for difficult separations.
Herein, the attributes of these phases are described and compared with the same phases on a high purity type B silica, and with conventional C18 Type B phases.
Selectivity of Type A and B C18 phases
In order to demonstrate the selectivity differences between the two types of silica, a range of uncapped trifunctional C18 phases have been produced. They are based on a base-deactivated type A silica and a high purity Type B silica. The bonding uses a procedure which allows a similar high degree of reproducibility as seen with the base silicas. A range of Extended Polar Selectivity (EPS) bonded phases was thus produced with accurately controlled varying levels of phase loading over the range 0.25 to 2.6 µmole m-2.
The same bonding procedure was used for both silicas and not adjusted for slightly different surface areas. This resulted in slightly lower phase levels expressed as µmole m-2 for the pure phases.
These were tested using the Snyder/Dolan hydrophobic subtraction model protocol to determine the six parameters which predict reversed phase selectivity. These parameters are:
S Steric Accessibility (NB current S/D protocol uses steric resistance which is the inversion).
A Hydrogen bonding with basic sites
B Hydrogen bonding with acidic sites
C 2.8 Ionic interaction with basic sites at pH 2.8
C 7 ionic interactions with basic sites at pH 7.0
Type B Phases
In figure one, the bonding on the pure type B silica is depicted. Parameter H of the Snyder/Dolan hydrophobic subtraction model protocol follow a linear increase (r = 0.99) with increasing phase level. This is as expected as it represents the purely hydrophobic interaction would be expected to increase in direct proportion to the level of C18 phase. The lack of scatter on the line is an indication of the degree of reproducibility.
There is a small decrease in steric accessibility (S) with increasing phase level as would be expected due to pore infill by the C18 phase. The hydrogen bonding of bases (A) shows a slight drop with the initial bondings but levels out. This probably results from the trifunctional phase. As silanols are lost on the silica due to bonding, additional silanols are available on the silane. Overall the effect is a relatively constant hydrogen bonding capacity for bases. Against this, the hydrogen bonding of acids (A) is low and trends barely observable.
There is a linear decrease in ion exchange interaction (C 2.8) as the phase level increases. One possible interpretation is that the ion exchange potential arises from the silica surface silanols which are rendered less accessible and reduced in number as the phase level increases. As the A term is relatively constant, it would appear that the additional silanols from the silane are not as acidic as the surface ones. Steric accessibility may play a part but the relatively constant A term would suggest that molecules are not unduly hindered as they appear to be able to hydrogen bond. The ion exchange potential at neutral pH (C 7) is surprisingly level. This implies that both residual silanols on the surface and additional silanes from the silane are equally ionised and accessible.
Type A Phases
The bonded Type A silicas show the following trends (Fig. 2): H, S, A and B show very similar trends to those seen on the pure silica. The H term gradients show similar values (0.21 vs.0.23) while the S term gradients show less steric accessibility for the Type A phases (-0.67 vs. -0.045). This may be due to the different pore diameters (100 A vs 120 A).
The two C terms however show distinctly different behaviour compared to the pure phases. Both C terms show increasing acidity with phase levels.
It is likely that the surface acidity of Type A silicas arises from surface and near surface high valence metals (Fe, Al, Ti) which withdraw electrons from nearby silanols increasing their acidity. At low bonding densities, excess surface silanols hydrogen bond with both surface metals and adjacent silanols decreasing the silanol acidity. As the bonding level rises, the number of surface silanols decreases, which increases the likelihood that a silanol will be isolated. If, in addition, the silanol is near a metal atom, the silanol acidity will be increased to a greater extent leading to increased ion exchange potential with bases.
The Selectivity Matrix
A cursory glance at the graphs shows that apart from the lowest level of bonding, neither of the silicas at any bonding level show even a slight similarity in selectivity. For a comparative point, an ideal fully bonded Type B silica which represents most phases in common use, would have an H value of 1.00, whilst all other parameters have a value of 0.00.
The presented data show that not only do the phases have extremely different selectivities compared to the standard C18 phases but that the ones based on Type A silicas are required to give the full gamut of selectivities.
The negative aspect of Type A based reversed phase media has been the peak shape with bases. However, the bonding used of these deactivated type A silicas allows reasonable peak shape with bases as can be seen in Fig. 3 which shows how the selectivity of amitriptyline varies with phase level on the studied phases.
Johannes Maisch1, Ian Chappell2
1Dr. A. Maisch HPLC GmbH, Ammerbruch, Germany
2Exmere Ltd., Carnforth, United Kingdom
Dr. Johannes Maisch
Dr. A. Maisch HPLC GmbH