Chiral Separation of Silanes by Use of Capillary Zone Electrophoresis

Connor Radecki

Summary

Capillary zone electrophoresis was used to separate two different chiral silanes.  By using sodium deoxycholate as a chiral selector in a TRIS/methanol buffer, successful separations were acquired for both silanes in similar conditions.  The two silanes, Silane 1 and Silane 2, can be seen below in Figure 1.  Silane 1 was dissolved in a 25 mM TRIS methanol solution, and run with a 100 mM sodium deoxycholate, 25 mM TRIS methanol buffer, resulting in a separation with a resolution of 1.38.  Silane 2 was also dissolved in a 25 mM TRIS methanol solution, but was run with a 70 mM sodium deoxycholate, 25 mM TRIS methanol buffer.  This separation was also successful, with a resolution of 1.49.

Introduction

A silane is any compound that is centered around a tetrahedral silicon atom.  Silanes have great importance in the chemistry field, due mainly to their use in stereoselective synthesis and in organometallic chemistry.  More recently, they are of great interest in the biochemical field, as they show abilities to be used in drug delivery of anticancer prodrugs.^[1]  Two silanes were analyzed in this project, with the ultimate goal of determining the diastereomeric ratio of both compounds.  This information could help in future synthesis and utilization of the silanes.  The structures of these compounds can be seen in Figure 1.

Figure 1. Molecular Structures of Silanes Investigated

The two silanes analyzed were very similar in structure, with the only variance being that Silane 1 contained an n-butyl group, whereas Silane 2 contained a napthyl group.  Therefore, it was determined that the conditions required for chiral separation would likely be very similar.  

The separation itself was done via capillary electrophoresis.  Capillary electrophoresis uses long and narrow (typically 20 to 200 micrometers in diameter) capillarys to separate both large and small molecules.  The separations are performed with the use of high voltages across the capillary, invoking a flow through the capillary, similar to the flow caused by the flow meter in high performance liquid chromatography.  There are two main causes of flow, the electrophoretic flow and the electroosmotic flow.  

The electrophoretic flow is the flow of the individual samples within the capillary.  Since there is a voltage difference applied across the capillary, one end of the capillary essentially becomes a cathode, and the other end becomes an anode.  This means that the charge of the sample plays a major role in the flow.  A molecule that is positively charged will be attracted to the cathode, and want to travel in that direction through the capillary.  A molecule that is negatively charged will want to travel in the opposite direction, towards the anode.  Another factor that affects the flow of the sample is the size and shape of the molecule.  A small molecule, such as a sodium ion, will be able to flow through the capillary quickly, due to a low size-to-charge ratio.  A larger and more bulky molecule, such as arginine, will travel much slower, due to a much higher size-to-charge ratio.  Because of this, arginine would likely take longer to reach the cathode than the sodium ion.  

The electroosmotic flow is the bulk flow of all of the sample, buffer, and solvent within the capillary.  It is similar to the flow in HPLC, in which it is a constant flow, unaffected by molecule charge or size.  In capillary electrophoresis, the capillaries used are often fused silica, which has a very low pI, around 1.5.  Because of this, the inner wall of the capillary becomes negatively charged.  Any positively charged ions within buffer are attracted to the negatively charged wall, and create a layer of positively charged ions surrounding the sample.  As these positively charged ions then travel towards the cathode, they “drag along” the rest of the sample, creating a bulk flow.

In order to create separation between the enantiomeric silanes, various cyclodextrins were used, as well as sodium deoxycholate.  Cyclodextrins are rings made up of glucose monomers, most often consisting of six, seven, or eight monomers, referred to as alpha-, beta-, and gamma- cyclodextrin, respectively.  The structure of a beta-cyclodextrin can be seen in Figure 2.

Figure 2. Structure of Beta-cyclodextrin

The cyclodextrin rings create a bucket-like shape into which the sample molecule fits.  However, part of the molecule is still exposed to the buffer in the capillary.  Since the molecules are enantiomers, the part of the sample molecule that is exposed to the buffer is slightly different in symmetry, and therefore will react with the buffer differently.  It is this variation in buffer reaction that ultimately allows for separation.  Use of sodium deoxycholate is very similar in the manner of separation, with the slight difference that the deoxycholate itself is not a bucket-like shape, but instead forms micelles, which produce very similar effects.  In 1993, Michel Neilen researched cyclodextrin’s ability to separate many different basic drug enantiomers^[2].   Through extensive testing of 10 different drugs, all of which were racemic mixtures, Neilen’s resulting peak resolutions were between 1.12 and 3.04.  Overall, the results showed that capillary electrophoresis did a very good job of separating enantiomers using cyclodextrins.  

Experimental Section

Apparatus

A Beckman Coulter (Fullerton, CA) Model 100-240 V, 8 A, 50/60 Hz P/ACE System MDQ Capillary Electrophoresis system was used, equipped with a variable wavelength UV absorbance diode array detector module.  The capillary zone electrophoresis was performed in a 30 cm fused silica Polymicro capillary, with an outer dimension of 364 μm and an inner dimension of 20 μm.  The injection time was 5.0 with a injection pressure of 0.5 psi.  

Reagents

Tris(hydroxymethyl)aminomethane (TRIS) and sodium lauryl sulfate, used in the stationary buffer phase, were obtained from Spectrum (New Brunswick, NJ).  Sodium deoxycholate and (2-hydroxypropyl)-β-cyclodextrin, both used as chiral selectors, were obtained from Sigma-Aldrich (St. Louis, MO).  Silane 1 and Silane 2 were synthesized by the Dr. Winchester research of Grand Valley State University (Allendale, MI).  

Procedures

-Buffer Preparation: A solution of Tris(hydroxymethyl)aminomethane in methanol was made in a 20 mL volumetric flask.  The solution required shaking and sonication to dissolve completely.  5 mL of this solution was then set aside, to be used for sample preparation.  Sodium deoxycholate was then added to the 15 mL of buffer, starting at 10 mM, and working up in 10 mM steps until separation was achieved.  

-Sample Preparation: The 5 mL of buffer set aside in a small sample tube was used to create a sample with an identical TRIS to methanol ratio as the buffer.  Using a lab spatula, a small scoop of the silane was added to the sample tube of buffer, and dissolved into it via shaking and sonication.  For some separations, a drop of dilute dimethyl sulfoxide was added as an electroosmotic flow detector.

-Running the separation: The 30 cm fused-silica capillary was inserted into the capillary electrophoresis system, and rinsed numerous times with methanol and dilute sodium hydroxide.  A 2 mL glass vial from the capillary electrophoresis instrument was filled with the sample, and two 2 mL glass vials were filled with the buffer solution.  In order to fill the capillary with the buffer, a one minute capillary rinse was done, flowing from the buffer to an empty vial in the instrument.  The voltage flow was set to 10 kV, and the injection of the sample was set for 0.5 seconds at 0.5 psi.  The run was then started, and allowed to run until at least 5 minutes past the last sample peak.  

Results

Separations were successfully obtained for both Silane 1 as well as Silane 2.  For Silane 1, the separation occurred with a sodium deoxycholate concentration of 100 mM in methanol.  The absorbance vs time graph of the separation of Silane 1 can be seen in Figure 3, and a graph of absorbance across various wavelengths vs time of Silane 1 can be seen in Figure 4.  Overall, it took about 9.6 minutes for the complete separation.  

Figure 3. Absorbance vs Time of Silane 1 at Specific Wavelength

Figure 4. Absorbance Across Range of Wavelengths of Silane 1

This separation had a resolution of 1.38, and was able to be replicated when run a second time.  We were able to determine this was a successful separation by comparing the absorbance pattern of the two peaks.  From previous runs consisting of just the silane in methanol, it was determined that Silane 2 absorbs most around 195 nm, and has a second, lesser absorbance area at around 220 nm.  This absorbance pattern matches the two peaks seen in Figure 4, which confirms that both peaks belong to Silane 2.  Also, it was known that the sample not racemic, but instead an uneven mixture of the two enantiomers, which can also be seen in Figure 3.

A separation was successfully obtained for Silane 2 as well, under very similar conditions.  For the separation of Silane 2, a concentration of only 70 mM sodium deoxycholate in methanol was required for the separation.  The absorbance vs time graph at a specific wavelength can be seen in Figure 5, and an absorbance across a range of wavelengths vs time can be seen in Figure 6.

Figure 5. Absorbance vs Time of Silane 2

Figure 6. Absorbance Across Range of Wavelengths of Silane 1

This separation had a resolution of 1.49, and was replicated when run a second time.  From previous runs of the silane in methanol, it was found that Silane 2 had three absorbance groupings, a large one at 190 nm, another large one at 220 nm, and a smaller third grouping down at about 280 nm.  This data allows for the conclusion that the separation seen in Figure 6 is in fact a separation of Silane 2.  When synthesized, it was speculated that this compound was a ratio of roughly 3:1, which matches up roughly with the peak sizes in Figure 5, further allowing for the conclusion that the peaks belong to both enantiomers of Silane 2.

In order to determine what concentration of sodium deoxycholate corresponded to the maximum bonding between the deoxycholate and the silane, a mobility plot had to be made.  The equation for mobility is,

                                              (1)

Where  stands for mobility, stands for the length of the capillary, stands for the length of the detector, stands for the voltage, stands for the time it took for the electroosmotic flow peak to appear, and stands for how long it took for the sample peak to come out.  A plot of concentration vs mobility should look like an exponential decay, since each increase in concentration causes a smaller change in bonding effect.  A table of the mobilities for the separation of Silane 1 and Silane 2 can be seen in Table 1.

Table 1. Mobilities of Silane 1 and Silane 2 vs Concentration of Sodium Deoxycholate

This data was then graphed, to visually show that the concentration of sodium deoxycholate used was at or near the maximum efficiency.  These graphs can be seen in Figure 7 for Silane 1, and Figure 8 for Silane 2.

Figure 7. Mobility vs Concentration of Sodium Deoxycholate of Silane 1

Figure 8. Mobility vs Concentration of Sodium Deoxycholate of Silane 2

Discussion

The best separation for both Silane 1 and Silane 2 occurred at sodium deoxycholate concentrations of 70 to 100 mM.  These separations occurred when using methanol solutions, and did not occur under the same concentrations but in water, or a water-methanol 50/50 mix.  One major issue in the use of water in the solutions was lack of solubility of the silanes.  Both silanes showed very limited solubility in water, and even water/methanol mixtures.  Because of this, many separation runs with water showed very small sample peaks, that were indiscernible as the peaks widened with addition of sodium deoxycholate.  

Prior to the use of sodium deoxycholate as a chiral selector, various cyclodextrins were used in attempt to separate the silanes.  -cyclodextrin was used first, in a TRIS buffer in water.  Following a similar ramping of concentration as sodium deoxycholate, the concentration of the -cyclodextrin was increased to 100 mM.  Although peak widening had occurred as the concentration was increased, a separation never ensued.  When mobility calculations were done, it was found that increasing the concentration past 100 mM would have very little effect.  A similar process was followed with hydroxypropyl cyclodextrin, dissolved in a TRIS buffer of a water-methanol mix.  This procedure showed similar peak widening, but never reached a separation.  Many other procedures were attempted, such as -cyclodextrin in water, the addition of sodium lauryl sulfate to introduce micelles, and methyl--cyclodextrin in water.  

Overall, the use of sodium deoxycholate proved to be more suited for separating the silanes than any cyclodextrins.  This could be due to a number of factors.  It is possible that the orientations the silanes took when fitting “into” the cyclodextrins were very similar, and thus reacted nearly identical when introduced to the voltage difference.  When the silanes were separated with the deoxycholate, however, the orientation of the micelles when surrounding the silanes must have differed enough to allow for separation when introduced to the voltage difference.  It is also possible that the silanes bound better with the deoxycholate than with the cyclodextrins.  Better binding would mean that increasing the concentrations would have a greater impact in mobility, thus allowing for better separations.

References

Xu, Li-Wen, Li Li, Guo-Qiao Lai, and Jian-Xiong Jiang. "The Recent Synthesis and Application

of Silicon-stereogenic Silanes: A Renewed and Significant Challenge in Asymmetric Synthesis." Chem. Soc. Rev. 40.3 (2011): 1777-790. Web.

Nielen, Michel W. F. "Chiral Separation of Basic Drugs Using Cyclodextrin-modified Capillary

Zone Electrophoresis." Analytical Chemistry Anal. Chem. 65.7 (1993): 885-93. Web.