Results and Discussion
There are a number of ways to develop a chromatographic separation of a given set of compounds. One way is to graph properties
such as retention or resolution as a function of a systematically controlled variable like mobile-phase pH. A resulting trend
may then show an optimal value for that parameter. It is suggested that this method should be avoided for two reasons. First,
it is an ad hoc investigation; in this approach, no hypothesis is proposed and tested. Only after a trend is observed is it
apparent what the analyst was looking for. No understanding of chromatographic theory is used to arrive at the optimum value,
but rather only a series of trial-and-error experiments. Second, the large number of runs required to generate these plots
uses unnecessary time and solvents.
In contrast, the method development process described in this study is an iterative process. Based on what is known about
how different variables affect retention in reversed-phase or ANP chromatography, the optimal conditions are predicted beforehand.
After the data is generated, the analyst compares the results to the predictions made in the hypothesis. As with most scientific
experiments, the amount of agreement between the obtained and predicted results leads to a second, more refined hypothesis
from which to construct additional experiments. This process not only quickly leads to the desired results for the separation
under study, but also to a more complete understanding of the chromatographic retention mechanism at work.
Reversed-phase chromatography was investigated first because of its widespread use in HPLC methodology. Because the retention
is based on hydrophobicity and the analytes are very polar, it was predicted that very high water content would be required
despite the known pitfalls of this combination. Thus, a 95% aqueous starting condition was selected for the gradient. A C18
stationary phase was chosen because of its strong hydrophobic character. A gradient ramp to 50% aqueous was sufficient to
elute the hydrophilic analytes. For the mobile phase additive, 0.1% formic acid was chosen to keep ascorbic acid neutral and
therefore more amenable to reversed phase retention. The data using the step 1 gradient with the Bidentate C18 column is shown
in Figure 2.
Figure 2: Individual standard injections using step 1 with the Bidentate C18 column. Standard injections are a) ascorbic acid,
b) riboflavin, c) pyridoxine, and d) thiamine.
Riboflavin was well retained but ascorbic acid, pyridoxine, and thiamine all exhibited minimal retention even at 95% aqueous
content in the mobile phase. The lack of retention for these compounds under highly aqueous conditions suggested that further
attempts in reversed phase were not likely to be successful. Higher retention could possibly be accomplished with the use
of ion pair agents such as alkyl sulfonates. However, this would make the method incompatible with MS detection. This data
is what prompted investigations in the ANP mode. The analytes could then be retained based on their polarity, thereby achieving
our method goals.
Using the low carbon-bonded silica hydride column, a 0.1% (v/v) formic acid mobile-phase additive was investigated. Because
thiamine has a permanent positive charge, it is expected to be highly susceptible to undesirable electrostatic interactions
with the very few residual silanol moieties remaining on the stationary phase surface. Operating under acidic conditions neutralizes
these residual silanol groups (8). Therefore, it was predicted that an acidic mobile phase would be required to avoid these
detrimental effects. A disadvantage of this mobile phase is that ascorbic acid will be neutral and therefore less hydrophilic
under these conditions; thus, a lower retention time would be expected than with use of an additive such as ammonium formate
or ammonium acetate.
Figure 3: Individual standard injections using step 2 with the Diamond Hydride column. Standard injections are (a) ascorbic
acid, (b) riboflavin, (c) pyridoxine, and (d) thiamine.
The ANP gradient separation using 0.1% formic acid (step 2) for individual standards is displayed in Figure 3. The data is
a vast improvement compared to the reversed-phase approach in terms of both retention and selectivity. However, ascorbic acid
shows low retention even with 95% starting organic content in the mobile phase. Therefore, the method should be modified in
a manner that leads to stronger ascorbic acid retention. Because the mobile-phase composition already starts out with low
solvent strength conditions, optimization of the gradient is not warranted at this stage. A better approach would be to operate
at a mobile-phase pH in which ascorbic acid is fully ionized to increase its hydrophilicity. The pH of the mobile phase should
be at least 1 pH unit above the pK
of ascorbic acid to ensure that it is fully ionized as ascorbate.