Understanding Separations in HILIC Chromatography: Impact of high Organic on Solute Descriptors David S. Bell,Hugh Cramer and Craig R. Aurand Supelco/Sigma-Aldrich 595 North Harrison Road Bellefonte, Pennsylvania 16823 USA

Abstract Interest in chromatography using aqueous-organic mobile phases high in organic content (aqueous normal-phase, ANP, HILIC) has continued to build in recent years.[1,2] In this mode of chromatography analyte retention increases monotonically with an increase in the organic component of the mobile phase. In previous studies, significant contribution of stationary phase chemistry toward the manipulation of retention and selectivity in ANP has been demonstrated.[3] The aim of this continuing study was to further enhance our knowledge of mechanisms of retention that dominate in this in this interesting mode of chromatography.

Solute descriptors important for chromatographic processes such as ionization constants (pKa/pKb) are often only available as measured in aqueous environments. When operating in HILIC mode, one deviates significantly from the aqueous environment and thus from these aqueous-based descriptors. It is therefore important to understand the state of the analytes and stationary phases in the HILIC environment. In this study, the influence of high organic content on basic and acidic pKa values as well as mobile phase pH is explored. The impact of the variation is related to real chromatographic separations and useful general trends are developed.

Introduction A significant contribution of ion-exchange interactions is known to be prevalent in HILIC systems [1,2]. This is especially evident on bonded polar phases, such as cyano and pentafluorophenyl phases Where ion-exchange occurs, the ionization state of both the analyte and the stationary phase surface can be used to predict and manipulate retention and selectivity Often, however, the descriptors (pKa, pH values) that are available to the analyst are based on aqueous measurements and do not necessarily reflect the ionization state as one moves into high organic mobile phases One must consider and understand the impact the organic modifier has on these parameters in order to efficiently and effectively develop methods in

Figure 1 shows the measurement of the effective pH (pH measured after the addition of the organic modifier - acetonitrile) for an ammonium acetate buffer system The measurements demonstrate that the effective pH tends to be greater than the aqueous based measurement often used (and assumed to be true). For example, a buffer prepared at pH 7 (aqueous) exhibits an effective pH of greater than 8 in 90% acetonitrile. A buffer prepared at pH 4 has an effective pH of nearly 7 The extent of this variation with organic is dependent on the buffer/modifier employed

Figure 1: Effect of Acetonitrile on pH of Ammonium Acetate [4] The pH of the aqueous ammonium hydroxide solution was adjusted with acetic acid prior to the addition of acetonitrile (x-axis). Subsequent pH measurement was taken following the addition of acetonitrile. Each measurement utilized a glass electrode filled with saturated KCl calibrated using pH 4, pH 7 and pH 10 NIST standardized aqueous reference. Measurements were taken at 25ºC. 90.0% 75% 50% 32.5% 6

Figure 1: Effect of Acetonitrile on pH of Ammonium Acetate [4] A Note on Buffer pH The pH of the aqueous ammonium hydroxide solution was adjusted with acetic acid prior to the addition of acetonitrile ( ). Subsequent pH measurement was taken following the addition of acetonitrile ( ). Each measurement utilized a glass electrode filled with saturated KCl calibrated using pH 4, pH 7 and pH 10 NIST standardized aqueous reference. Measurements were taken at 25ºC. Triangle: 90.0% ACN, Square: 75% ACN, Diamond: 50% ACN, Circle: 32.5% ACN Demonstrates the need to understand how pH changes with the addition of organic – it is suggested to measure the pH following the addition of organic. In this experiment we used buffer that was unadjusted or at about a pH of 6.7 – meaning at 90% acetonitrile the effective pH is about 8

Analyte pKa values have also been shown to be impacted by the presence of organic modifiers Figure 2 shows the results of an NMR experiment conducted that explored the chemical shift of a proton near the ionizable group for amitriptyline in 90% acetonitrile. From data such as this, effective pKa values can be established for a variety of compounds. Table 1 shows the results for several basic pharmaceutical compounds. The data indicates that the effective pKa value for a basic analyte in 90% acetonitrile is approximately 1 pKa unit less than the aqueous-based value

Figure 2: Determination of pKa Values using 1H NMR [4] Amitriptyline Chemical Shift as a Function of pH at 90% Acetonitrile

Table 1: Determination of pKa Values using NMR [4] Amitriptyline Analyte Literature pKa pKa Correlation (R2) Amitriptyline 9.4 8.34 0.9923 Nortriptyline 9.7 8.92 0.9920 Diphenhydramine 9.0 8.33 0.9978 Verapamil 8.9 7.98 0.9976 Alprenolol 8.73 0.9855 % Acetonitrile pKa Correlation (R2) 25 9.32 0.9997 50 9.02 0.9996 75 8.88 0.9956 90 8.34 0.9923 pKa values for bases decrease with increasing acetonitrile At 90% each analyte exhibited a pKa value about 1 full pH unit less than the literature pKa value

Impact of pH and pKa Variation on Ion-Exchange In order for an ion-exchange interaction to take place, both the analyte and the stationary phase must posses opposite charges Taking only the aqueous-based values: Analyte pKa = 8.0 Mobile phase pH = 7 The degree of analyte ionization would be: 10 (pKa – pH) /(1 + 10 (pKa-pH) ) = 0.90 or 90% ionized In 90% acetonitrile, however: Analyte pKa ~ 7.0 Mobile phase pH ~ 8 The degree of analyte ionization would be only about 10% and thus much less apt to interact via ion-exchange It is thus extremely important to take the variation of both pH and pKa into account when developing HILIC or highly organic LC methods

Experimental Basic analytes ( See Figure 3) ranging in pKa values from approximately 7.5 to 10.5 were run on an pentafluorophenyl phase, known to exhibit ion-exchange character, as a function of percent acetonitrile at an aqueous-based pH of both 6.8 and 4.0 Retention was monitored and assessed based on the known impact of increasing acetonitrile on both analyte pKa and mobile phase pH values Calculated (aqueous-based) physical parameters are presented in Table 2

Conditions: Instrument: Waters 2690 coupled to a Waters/Micromass ZQ single quadrupole mass spectrometer via an electrospray interface operating in positive ion mode Column: Ascentis Express F5 (pentafluorophenylpropyl-bonded) 10 cm x 3.0 mm, 2.7 m particle size Mobile Phase A: either 10 mM ammonium acetate, pH 6.8 (unadjusted) or 10 mM ammonium acetate, pH to 4.0 with glacial acetic acid Mobile Phase B: acetonitrile Isocratic elution obtained by on-line mixing from 70% A to 10% A Flow Rate: 0.5 mL/min Temperature: 35ºC Sample Injection Volume: 2 mL – made in triplicate at each level

Figure 3: Analyte Structures

Table 2: Calculated Physical Parameters for Analytes [5] Compound Name pKa(MB) LogD(7.4) LogP MW Selegiline 7.53 2.31 2.68 187.28 Lidocaine 7.96 1.67 2.2 234.34 Mepivacaine 8.29 1.28 1.78 246.35 Procainamide 9.09 -0.56 1.32 235.33 Amphetamine 9.94 -0.65 1.79 135.21 Methamphetamine 10.38 -0.53 149.23

Results The retention data for the analytes as a function of percent acetonitrile at pH 6.8 and pH 4.0 are graphically presented in Figures 4 and 5, respectively The three analytes with the highest pKa values (>9), procainamide, amphetamine and methamphetamine all exhibit “U-shaped” retention at both pH 6.8 and pH 4. The upswing in retention at high organic in the “U-shape” retention curve has been shown to be a result of ion-exchange mechanisms using fluorinated phases. [2-4] The remaining weaker bases show a linear decrease in retention with increasing acetonitrile at pH 6.8 indicating a lack of ion-exchange under these conditions.

At pH 4. 0 the weaker bases show a complex behavior At pH 4.0 the weaker bases show a complex behavior. Curvature at lower organic percentage indicates some ion-exchange behavior, however at higher organic content, the retention dramatically drops off. The drop off is related to loss of degree of ionization as the effective pH of the mobile phase increases and effective pKa values of the analytes decrease with increasing acetonitrile. It is interesting to note that the drop off occurs at lower organic for the weakest base, selegiline, higher for the intermediate base, lidocaine, and even higher for the stronger of the weak bases, mepivacaine.

Figure 4: Retention as a Function of % Acetonitrile, pH 6.8

Figure 5: Retention as a Function of % Acetonitrile, pH 4.0

75% Acetonitrile 60% Acetonitrile 80% Acetonitrile Figure 6: Altering Retention and Selectivity through Manipulation of Ion-Exchange Component 75% Acetonitrile 60% Acetonitrile 80% Acetonitrile

Retention on fluorinated phases is dominated by both traditional reversed-phase mechanisms (partitioning) and by ion-exchange mechanisms. Through an understanding of how pKa and pH values vary as a function of organic, it is possible manipulate the ion-exchange component along with partitioning to produce the desired chromatography. Figure 6 shows how elution order of selegiline and two of its potential impurities, amphetamine and methamphetamine, can be altered with partitioning and ion-exchange mechanisms in mind. At 60% acetonitrile selegiline elutes after the amphetamines. Increasing the organic to 75% results in selegiline eluting between the two amphetamines, whereas 80% organic results in elution prior to the two stronger bases. Selegiline, the weaker thus less ionized base, retains primarily by partitioning whereas the amphetamines retain primarily by ion-exchange

Conclusions Polar phases such as pentaflourophenyl, cyano and bare silica exhibit ion-exchange behavior. Ion-exchange can be a powerful mechanism to separate analytes of differing degrees of ionization (pKa values), however a firm understanding of the origin of such values and how they change as a function of organic modifier is essential for efficient method development. The impact of the percentage of acetonitrile on both analyte pKa and mobile phase pH has been correlated with the retention of basic compounds on fluorinated phases. Compounds that would be expected to be ionized using aqueous-based values are shown to be poorly retained where ion-exchange mechanisms dominate (high acetonitrile percentages)

A convenient estimation (rule of thumb) can often be applied: The effective pKa of a base is approximately one pKa unit lower than the aqueous-based measurement in 90% acetonitrile This, combined with a measurement of the pH after addition of the organic, provides a more accurate determination of actual degree of ionization and thus ion-exchange behavior

References [1] W. Naidong, Journal of Chromatography B 796 (2003) 209. [2] D.S. Bell, Jones, A. Daniel, Journal of Chromatography A 1073 (2005) 99. [3] D.S. Bell, Brandes, Hillel K., in 30th International Symposium and Exhibit on High Performance Liquid Phase Separations and Related Techniques, San Francisco, California USA, 2006. [4] D.S. Bell, Solute Attributes and Molecular Interactions Contributing to Retention on a Fluorinated High-Performance Liquid Chromatography Stationary Phase, Thesis, The Pennsylvania State University, 2005 [5] ACD PhysChem, v. 12, Advanced Chemistry Development, Toronto, ON Canada