An Unexpected Observation Concerning the Effect of Anionic Additives on the Retention Behavior of Basic Drugs and Peptides in Reversed-Phase Liquid Chromatography

Abstract

Anionic species with ion pair forming ability are commonly used to enhance the retention and efficiency of basic analytes in RPLC separations. However, little is known about the interactions between organic mobile phase modifiers and such ion pairing anions. In this work, we measured the magnitude of the retention increase of basic drugs and peptides upon addition of strong inorganic ion pairing anions (e.g. perchlorate) as a function of the volume fraction of modifier in acidic water-acetonitrile mobile phases on two different stationary phases. We found that the increase in retention upon addition of various salts depended strongly on the eluent strength. In general, larger retention increases upon addition of the anion were observed at higher organic fractions. Regression of retention against the volume fraction of organic modifier indicated that the ion pair forming anions substantially decreased S values while only slightly changing ln k’_w values. The decrease in S is the major cause of the retention increase of basic drugs and peptides when such anions are added to the mobile phase.

1. INTRODUCTION

The separation of basic compounds by RPLC is of great importance in both the pharmaceutical analysis and proteomic research. Most RPLC columns are packed with silica-based alkyl silane bonded stationary phases. However, detrimental interactions between cationic analytes and surface silanols can occur which frequently lead to peak tailing and poor efficiency. Typically, mobile phases containing acids (e.g. phosphoric acid, formic acid or trifluoroacetic acid (TFA)) are used to suppress these interactions [1,2]. However, in acid the positive charge of the analyte can cause low retention and thus poor resolution. To mitigate this problem, anions are often added to the eluent to adjust the retention, efficiency and sample loading capacity of basic analytes [3-11].

The study of the retention mechanism of basic analytes in RPLC in the presence of very hydrophobic anionic additives such as alkyl sulfonates was pioneered by Horvath et al. [12], followed by further studies by Bidlingmeyer et al. [13], Knox et al. [14] as well as an extensive review by Weber and Cantwell [15]. Both the ion pairing and dynamic ion-exchange mechanisms were invoked to explain the increased retention. Unfortunately, the highly hydrophobic alkyl sulfonates have a tendency to “stick” to the stationary phase making recovery of column properties difficult. Recently, less hydrophobic anionic additives (e.g. CF₃COO^-, ClO₄^-, PF₆^-) have gained in popularity for use with basic compounds and their influence on their retention mechanism has been the subject of extensive studies [4,5,7,9,16].

In general, adding ion pairing anionic additives increases the retention of basic compounds. More hydrophobic anions such as PF₆^- cause the largest retention increase, followed by ClO₄^-, CF₃COO^-, Cl^-, HCOO^- and H₂PO₄^-. Roberts vaguely referred to the “Hofmeister effect” to explain this observation [9]. Kazakevich and coworkers attributed the retention increase due to addition of inorganic anions to a “chaotropic effect” [6]. Gritti and Guiochon studied the role of buffer (e.g. phosphate and acetate) on retention and overloading behavior of cationic drugs on different stationary phases [17-19]. They suggested that ion pair formation could explain their experimental results. It is clear that there was little agreement and measurements of the extent of ion pair formation were necessary to definitively settle this complex phenomenon.

Dai et al. determined ion pair formation constants between various basic drugs and different anions under conditions representative of the eluents used in RPLC through independent measurements of the effect of anions on the mobility of the cations of interest by capillary electrophoresis [20]. Their measurements confirmed that at pH 4-5 ion pair formation in the eluent is responsible for the large increases (2-5 folds) of retention of drugs upon addition of anions to the eluent. It is important to note that at typical anion concentrations (< 100 mM) less than 50% of the test bases were actually present as ion pairs in the mobile phase even with the strongest pairing agent (PF₆^-) used. At pH 2 slight but still significant adsorption of anionic additives to the stationary phase was also confirmed by their effect on the retention of anions such as Br^-; these adsorbed anionic species generate dynamic ion exchange sites for retaining the cationic analytes [4]. Thus, both ion pair formation and dynamic ion exchange, especially at lower pH, contribute to the overall retention increase of the basic drugs upon addition of anions. The relative contribution depends on the analyte and the operating conditions (e.g. mobile phase, especially the pH and stationary phase).

Due to both of the above mechanisms anions are also found to be extremely useful for improving the separation of proteins and peptides. For instance, TFA is routinely used to control the pH and improve the peak shape of peptides [21]. Hodges and coworkers have extensively studied the effect of different anionic additives on the retention and selectivity of peptides in gradient elution RPLC [3,10,11,22]. The ability of different anions to increase the retention of peptides in gradient elution agrees well with the results of retention increases for basic drugs in isocratic elution. They found that the retention increase depends strongly on the charge of the peptides and the concentration of the additive. For example, the addition of 50 mM sodium perchlorate to 10 mM phosphoric acid buffer increased the gradient retention times of peptides with +4 charges by up to 15 minutes [22].

A very important feature of anion ion pair forming additives, besides the increase in retention, is their ability to decrease the peak width of both basic drugs and biological molecules. In the absence of such strong ion pairers, wide peaks for cationic analytes are usually observed compared to neutral analytes on most stationary phases; this is mainly caused by the fact that typical RPLC stationary phases are overloaded with the positively charge analytes due to their poor loading capacity. McCalley studied the effect of buffer type, concentration and ionic strength at low pH on the efficiency of basic drugs [23,24] and peptides [25]. He showed that high ionic strengths and stronger ion pairing agents significantly increase the sample loading capacity of basic analytes. As a result, much better plate counts can be obtained when the same amount of cationic analyte is injected.

It is clear that addition of anionic additives is very beneficial for the separation of cationic analytes. Of particular importance is the ability of anionic additives to increase retention under both isocratic and gradient conditions. However, all previous work measured the magnitude of isocratic retention increases due to anionic additives at constant eluent strength (i.e. fixed organic modifier fraction in the eluent with different types and concentrations of additives). Thus, little is known about how the eluent strength itself affects the magnitude of the retention increase when anions are added to the eluent. Due to the electrostatic nature of ion-ion interactions, we should not overlook the fact the eluent composition could play an important additional role in RPLC by affecting the dielectric constant of the mobile phase [7].

The retention behavior of basic analytes inThe retention behavior of basic analytes in RPLC when the eluent strength is varied can be described, at least approximately, by linear solvent strength theory (LSST) [26]:

$$\ln k’=\ln k_w^{’}-S·\varphi$$ (1)

where ln k’ is the logarithmic retention, ln k’_w is the extrapolated retention in pure water, ϕ (i.e. %ACN) is the volume fraction of acetonitrile (ACN) in the mobile phase, and S is the slope of a plot of ln k’ vs. ϕ. Regression of ln k’ against ϕ using eq 1 gives the ln k’_w and S values for different analytes; these are important parameters for predicting the retention time and peak width in gradient elution separations. Since proteomic samples are invariably separated under gradient conditions, we studied the effect of ion pairing anionic additives on the retention behavior of basic drugs and peptides as a function of eluent strength using LSST. As will be shown later, the presence of the anion has only a small effect on ln k’_w, but it significantly decreases the S values of basic compounds (both low molecular weight drugs and higher molecular weight peptides); the decrease in S is clearly the main factor which drives the increase in retention of basic compounds.

2. EXPERIMENTAL

2.1. Materials and Reagents

All solutes were of reagent grade or better and were used without further purification. All cationic drugs and peptides were obtained from Sigma (St. Louis, MO, USA). Trifluoroacetic acid (99%) and formic acid (88%) were purchased from Aldrich (Milwaukee, WI, USA). Sodium perchlorate and sodium chloride were obtained from Fisher (Fair Lawn, NJ, USA) and Mallinckrodt (Hazelwood, MO, USA), respectively. HPLC water was obtained from a Barnsted Nanopure deionizing system (Dubuque, IA, USA) with an “organic-free” cartridge and a 0.2 μm filter. HPLC grade acetonitrile (ACN) was purchased from Sigma-Aldrich (Milwaukee, WI, USA). All pure solvents were filtered through a 0.45 μm nylon filtration apparatus (Lida Manufacturing Inc., Kenosha, WI) before use. All eluents were prepared by weight. These eluents were then used immediately after mixing without degassing.

2.2. HPLC Instrumentation and Columns

All chromatographic experiments were conducted using an HP 1090 HPLC controlled by version A.10.01 Chemstation software (Agilent Technologies, Palo Alto, CA). The instrument was equipped with a low pressure mixing chamber, autosampler, photodiode-array UV detector, an 8 μL flow cell detector with 6 mm path length and binary pump. Column temperature was controlled by using an eluent pre-heater and column heating jacket that were generous gifts from Systec Inc. (New Brighton, MN). The HP1090 block heater was by-passed by directly connecting the injection valve to the Systec eluent pre-heater with 10 cm of PEEK tubing (0.007” i.d.) to reduce the system dwell volume. The dwell volume was measured to be 0.30 mL [27]. The eluent exiting the HPLC column was cooled before entering the detector flow cell using the built-in HP1090 heat exchanger.

The 50 mm × 2.1 mm i.d. Zorbax SB-C18 column was a gift from Agilent Technologies (Wilmington, DE, USA). The particle size was 3.5 μm and the average pore diameter was 80 Å. The 50 mm × 4.6 mm i.d. Ace 5 C18 column was a gift from Mac-Mod Analytical, Inc. (Chadds Ford, PA, USA). The particle size was 5 μm and the average pore diameter was 100 Å.

2.3. Chromatographic Conditions

Chromatographic measurements were made at a flow rate of 0.40 mL/min on the SB-C18 column and at a flow rate of 1.00 mL/min on the Ace 5 C18 column. The detection of the drugs and peptides was set at 214 nm in 0.1% TFA containing mobile phases or 250 nm in 0.1% formic acid containing mobile phases. The detection of the nitrate ion was at 210 nm. The concentration of ion pairing anions was adjusted by adding the proper volume of 500 mM sodium perchlorate or sodium chloride stock solutions to the mobile phase. Drug samples were prepared in the mobile phase (about 0.2 mM) and the anionic probe sample (i.e. KNO₃) was prepared in water (about 0.2 mM). Typical injection volumes were 2 μL and 5 μL on the SB-C18 column and the Ace 5 C18 column, respectively. The column temperature was controlled at 40 °C. At each mobile phase composition, three injections were made to obtain replicate k’ values.

To obtain the gradient ln k’_w and S values, solvent A was 0.1% TFA (v/v) in H₂O and solvent B was 0.1% TFA (v/v) in 80:20 ACN:H₂O. The linear gradient profile was from 35% to 55% B in 5, 10, or 20 min for the basic drug mixture (Alprenolol, Desipramine, Amitriptyline and Perphenazine) and was 10% to 60%B in 15, 30, 60 min for the peptide mixture (see caption of Figure 7). The detailed procedure to obtain the gradient ln k’_w and S values for each solute using these training runs is described elsewhere [28].

Figure 7.

Comparison of (A) ln k’_w and (B) S values of ten peptides on SB-C18 in two different mobile phases. The ln k’_w and S values are obtained from three gradient training runs. Peptide: (1) Neurotensin fragment 1-8; (2) Phe-Phe; (3) LHRH; (4) Angiotensin II; (5) [Val⁵]-Angiotensin I; (6) Substance P; (7) Renin substrate; (8) Momany peptide; (9) Insulin chain B oxidized; (10) Melittin. Plot legends from left to right correspond to mobile phase containing: No anionic additive; 20 mM NaClO₄. All mobile phases contained 0.1% TFA. Other experimental conditions: 0.40 mL/min, gradient from 8% to 48% ACN in 15, 30 or 60 minutes, 40 °C, 214 nm.

2.4. Weighted Least Square Linear Regression

Since a logarithmic transformation of k’ is done when one regresses ln k’ vs. %ACN, the data points at the different eluent strengths must be properly weighted to account for the very different experimental errors of the different k’ values. Following the advice of Taylor [29] and Bevington [30] the data were weighted in proportion to k’². The intercept and slope, and their standard errors were calculated using the equations given in Ref. [29]. Clearly, the data points at the higher %ACN (i.e., smaller k’) are given less weight than those at lower %ACN. Therefore, low k’ data points are more poorly fitted to the line as will be shown later (see Figure 1, 3, and 5). Weighted least squares linear regression inherently gives smaller standard errors and narrower confidence intervals for both ln k’_w and S compared to normal linear regression (i.e., the equally weighted case).

Figure 1.

Retention of amitriptyline as a function of ACN volume fraction in three different mobile phases. Amount and type of anionic additive in the mobile phase: (●) None; (○) 20 mM NaClO₄; (▼) 40 mM NaClO₄. The stationary phase was Zorbax SB-C18 and all mobile phases contained 0.1% TFA. Other experimental conditions: 0.40 mL/min, 40 °C, 214 nm.

Figure 3.

Retention of amitriptyline as a function of ACN volume fraction in three different mobile phases. Amount and type of anionic additive in the mobile phase: (●) None; (○) 20 mM NaCl; (▼) 20 mM NaClO₄. The stationary phase was Zorbax SB-C18 and all mobile phases contained 0.1% formic acid. Other experimental conditions: 0.40 mL/min, 40 °C, 250 nm.

Figure 5.

Retention of amitriptyline as a function of ACN volume fraction in three different mobile phases. Amount and type of anionic additive in the mobile phase: (●) None; (○) 20 mM NaCl; (▼) 20 mM NaClO₄. The stationary phase was Ace 5 C18 and all mobile phases contained 0.1% formic acid. Other experimental conditions: 1.00 mL/min, 40 °C, 250 nm.

3. RESULTS AND DISCUSSION

3.1. Effect of Anionic Additives on Cationic Drugs on SB-C18 in 0.1% TFA

The retentions of four prototypical basic drugs in 0.1% TFA on Zorbax SB-C18 were measured at different mobile phase compositions. Figure 1 shows a plot of ln k’ as a function of volume fraction of acetonitrile for amitriptyline in the presence of 0, 20, and 40 mM NaClO₄. These concentrations were deliberately selected to obtain significantly different retentions under each condition [4,7]. It is clear that the retention of amitriptyline increases as the concentration of NaClO₄ is increased. The increase was larger from 0 to 20 mM NaClO₄ than from 20 to 40 mM NaClO₄; this is consistent with the results of others [4]. It should be noted that both perchlorate and trifluoroacetate in the mobile phase contributed to the observed retention increase. However, surprisingly, the retention of amitriptyline (as well as all other drugs) is less dependent on the NaClO₄ concentration at the lower %ACN than at the higher %ACN.

To better understand this result, we measured the ln k’_w and S values of the four drugs as a function of NaClO₄ concentration (see Figures 2A and 2B). First, it was quite surprising to see that the ln k’_w values of all four drugs decreased slightly and certainly did not increase upon addition of NaClO₄. This suggests that the addition of NaClO₄ would not enhance their retention from pure water to the stationary phase. However, this prediction is not certain since plots of ln k’ vs. %ACN tend to become quite non-linear at low %ACN. Indeed, inspection and detailed analysis of the data fitting of our plots indicates curvature. Nonetheless, it is very well known that the extraction of hydrophobic cations from water into low polarity solvents such as c-hexane and methylene chloride is greatly enhanced by changing the anion from chloride to perchlorate and other more hydrophobic anions [31,32].

Figure 2.

Comparison of (A) ln k’_w and (B) S values of four basic drugs in three different mobile phases. The ln k’_w and S values are obtained from weighted least square linear regression of ln k’ vs. %ACN. Solute: (1) Alprenolol; (2) Desipramine; (3) Amitriptyline; (4) Perphenazine. Plot legends from left to right correspond to mobile phase containing: No anionic additive; 20 mM NaClO₄; 40 mM NaClO₄. All conditions are the same as in Figure 1.

It is likely that the extrapolated value of ln k’_w is not a valid estimate of partitioning between water and the pure stationary phase. Rather it is well known that the amount of acetonitrile in the stationary phase increases rapidly as the amount in the mobile phase is first increased [33]. The ln k’_w is a measure of partitioning between pure water and a hypothetical acetonitrile equilibrated phase.

More surprisingly, the S values decreased substantially for all drugs when 20 mM NaClO₄ was added. Increasing the concentration of NaClO₄ to 40 mM further decreased the S values, although only slightly. Clearly, the addition of 20 mM NaClO₄ tends to have a greater effect on the retention of basic drugs in ACN richer mobile phases.

3.2. Effect of Anionic Additives on Cationic Drugs on SB-C18 in 0.1% Formic Acid

Clearly, the addition of an ion pairing anionic additive such as perchlorate to the mobile phase has changed the retention behavior of the basic drugs. In the previous experiment, the initial mobile phase contained 0.1% TFA; trifluoroacetate is known to form ion pairs with positively charged analytes [20] although it is generally considered a weaker ion pairing agent than perchlorate. To better understand the effect of ion pairing, we added 20 mM NaClO₄ to mobile phases containing 0.1% formic acid. Formate is a much weaker ion pair former with basic drugs than is trifluoroacetate. We hoped that the effect of NaClO₄ on the retention behavior of basic drugs would be clearer. We also tested the addition of 20 mM NaCl to 0.1% formic acid to compare the results to those with NaClO₄.

Figure 3 shows the retention of amitriptyline as a function of ACN volume fraction on in three different mobile phases. Clearly, 20 mM NaClO₄ causes the retention to increase more than did 20 mM NaCl compared to an eluent without any added salt. According to our previous studies [20], whereas chloride does form ion pairs with basic drugs, its counter-ion, i.e. sodium ion, decreases the contribution to retention from ion-exchange of the cationic analyte with surface silanols. Under acidic conditions (e.g. 0.1% formic acid) the retention contribution from ion-exchange is minimal which means that the addition of 20 mM NaCl increased retention through ion pairing of chloride with the cationic drugs. The stronger ion pairing ability of perchlorate led to a greater retention increase than did chloride at the same concentration.

The ln k’ values were fitted against eluent strength to obtain the ln k’_w and S values for desipramine and amitriptyline (see Figure 4 and Table 1). In contrast, addition of 20 mM NaCl and 20 mM NaClO₄ caused very slight increases in the ln k’_w values of the two drugs. On the other hand, 20 mM NaClO₄ greatly decreased the S values of both drugs while 20 mM NaCl only marginally decreased the S values. Since the ionic strengths in 20 mM NaCl and 20 mM NaClO₄ containing mobile phases are the same, the smaller S values in 20 mM NaClO₄ clearly must be due to the greater ion pairing ability of perchlorate and are not due to suppression of an ion exchange effect with ionized surface silanols.

Figure 4.

Comparison of (A) ln k’_w and (B) S values of two basic drugs in three different mobile phases. The ln k’_w and S values are obtained from weighted least square linear regression of ln k’ vs. %ACN. Plot legends from left to right correspond to mobile phase containing: No anionic additive; 20 mM NaCl; 20 mM NaClO₄. All conditions are the same as in Figure 3.

Table 1.

Effect of anionic additive on the ln k’_w and S values of two basic drugs on two stationary phases in mobile phases containing 0.1% formic acid

Condition	SB-C18a			Ace C18b
	1c	2d	3e	1c	2d	3e
Desipramine
lnk’w	6.72	6.87	6.95	6.80	6.65	6.21
SE (lnk’w)	0.01	0.01	0.02	0.02	0.02	0.02
S	18.06	17.69	14.93	17.38	15.81	12.34
SE (S)	0.06	0.04	0.06	0.09	0.07	0.07
Amitriptyline
lnk’w	7.20	7.34	7.44	7.24	7.10	6.67
SE (lnk’w)	0.01	0.01	0.01	0.02	0.01	0.02
S	18.47	18.11	15.39	17.77	16.24	12.79
SE (S)	0.04	0.03	0.05	0.07	0.05	0.05

^aChromatographic conditions on SB-C18 are given in Figure 3

^bChromatographic conditions on Ace C18 are given in Figure 5

^cThe mobile phase was 0.1% formic acid without any anionic additive

^dThe mobile phase was 0.1% formic acid with 20 mM NaCl

^eThe mobile phase was 0.1% formic acid with 20 mM NaClO₄

3.3. Effect of Anionic Additives on Cationic Drugs on Ace C18 in 0.1% Formic Acid

Column silanophilicity can affect the ion-exchange contribution from surface silanols to retention of basic compounds [34]. Therefore, we studied the effect of changing from an SB-C18 phase to the less silanophilic Ace C18 using the same mobile phase conditions as in Section 3.2 [34,35]. Figure 5 shows plots of ln k’ of amitriptyline vs. volume fraction of ACN in 0.1% formic acid with no salt, 20 mM NaCl and 20 mM NaClO₄. The addition of 20 mM NaCl caused a slightly greater increase in retention compared to the results on SB-C18 (see Figure 3). This is consistent with the fact that there is a smaller retention contribution from ion-exchange with surface silanols on Ace C18 and this ion-exchange retention decreases as more sodium ion is added. The ln k’_w and S values under these conditions are summarized in Figure 6 and Table 1. It is obvious that the S values of both desipramine and amitriptyline decreased upon addition of NaCl and NaClO₄. Furthermore, the ln k’_w values of both drugs decreased upon addition of 20 mM NaClO₄.

Figure 6.

Comparison of (A) ln k’_w and (B) S values of two basic drugs in three different mobile phases. The ln k’_w and S values are obtained from weighted least square linear regression of ln k’ vs. %ACN. Plot legends: same as Figure 4. All conditions are the same as in Figure 5.

3.4. Comparison of LSST Values of Cationic Drugs from Isocratic and Gradient Experiments

The ln k’_w and S values of the basic drugs can also be measured by doing several “training” runs under gradient conditions. In general, these values determined from gradient runs are simply the best values that satisfy certain equations for predicting gradient retention times. The physical meaning of ln k’_w and S values based on gradient elution is not as clearly defined as those from isocratic experiments. To test whether the gradient values agree with the isocratic values, we obtained ln k’_w and S values of the four basic drugs through gradient experiments on the SB-C18 column in 0.1% TFA containing mobile phase.

The gradient elution based values were compared to the isocratic values obtained in Section 3.1 (see Table 2). Generally, the ln k’_w and S from gradient experiments are in good agreement with the isocratically based values. It should be noted that the standard errors of gradient values are larger than those of isocratic values. This is mainly caused by the fact that the isocratic values were obtained by weighted linear regression whereas such a model is not available for fitting the gradient values. Overall, the trends in gradient based ln k’_w and S after addition of 20 mM NaClO₄ agree with the isocratic values: the addition of 20 mM NaClO₄ caused a slight decrease in ln k’_w but a big decrease in S values. Based on the good agreement between the isocratic and gradient results with the small drug solutes we believe that it is valid to use gradient based values of ln k’_w and S for peptides (see next Section).

Table 2.

Comparison of the ln k’_w and S values of four basic drugs on SB-C18 in mobile phases containing 0.1% TFA from isocratic and gradient experiments

[ NaClO4]	Isocratica			Gradientb
	0 mM	20 mM	40 mM	0 mM	20 mM	40 mM
Alprenolol
lnk’w	5.47	5.15	5.03	5.62	5.19	5.48
SE (lnk’w)	0.04	0.05	0.03	0.09	0.15	0.08
S	14.96	12.48	11.63	15.23	12.38	12.88
SE (S)	0.16	0.15	0.08	0.38	0.54	0.29
Desipramine
lnk’w	8.06	7.80	7.63	7.93	7.79	7.90
SE (lnk’w)	0.04	0.04	0.03	0.07	0.06	0.08
S	21.99	18.74	17.45	21.21	18.54	18.08
SE (S)	0.17	0.15	0.10	0.27	0.21	0.27
Amitriptyline
lnk’w	7.34	6.94	6.73	7.17	6.90	6.94
SE (lnk’w)	0.02	0.02	0.02	0.07	0.06	0.07
S	18.09	15.24	14.13	17.17	14.94	14.59
SE (S)	0.08	0.08	0.05	0.28	0.21	0.23
Perphenazine
lnk’w	7.81	7.42	7.22	7.47	7.20	7.28
SE (lnk’w)	0.02	0.02	0.01	0.10	0.08	0.08
S	18.54	15.71	14.60	17.02	14.82	14.61
SE (S)	0.06	0.06	0.04	0.36	0.25	0.27

^aIsocratic chromatographic conditions are given in Figure 1

^bGradient chromatographic conditions: 0.40 mL/min, gradient from 28% to 44% ACN in 5, 10, or 20 minutes, 40 °C, 214 nm

3.5. Effect of Anionic Additives on Peptides on SB-C18 in 0.1% TFA

Hydrophobic anionic additives are very useful in adjusting the retention and improving the efficiency of peptide separations [10,11,25]. However, due to the much larger S values of peptides, the measurement of ln k’_w and S under isocratic conditions for peptides is extremely difficult. On the other hand, gradient experiments under different gradient conditions provide a convenient alternative method of determining these parameters [26].

The ln k’_w and S values of ten commercially available peptides were measured on the SB-C18 column in 0.1% TFA containing mobile phases using gradient training runs. Figure 7 shows the comparison of the ln k’_w and S values before and after the addition of 20 mM NaClO₄. It is evident that for the peptides the addition of 20 mM NaClO₄ does not affect the ln k’_w values whereas the S values decrease significantly. This trend agrees very well with the results for the basic drugs under various conditions. Therefore, the decrease in S upon addition of ion pair forming anions is likely general for all cationic analytes.

In gradient elution, a very important parameter that describes solute retention is the effective retention factor (k*), which is given by:

$$k\ast =\frac{1}{0.5b+(1/k_0^{’})}$$ (2)

where k’₀ is the retention factor of the solute in the initial eluent of the gradient [26]. The b in eq 2 is the gradient steepness and is defined as:

$$b=\frac{S·\mathrm{\Delta }\varphi ·V_m}{F·t_G}$$ (3)

where F is the flow rate (mL/min), t_G is the gradient time, V_m is the column dead volume (mL1) and Δϕ is the change in eluent strength during the gradient. When the initial eluent is weak (i.e. low %ACN), k’₀ is very large (especially for peptides with big S values) and eq 2 simplifies to:

$$k\ast =2/b=2F·t_G/(S·\mathrm{\Delta }\varphi ·V_m)$$ (4)

As we have shown when anions are added to the eluent, ln k’_w values are minimally affected whereas the S values are decreased significantly. Therefore, according to eq 4, it is clear that the significant decrease in S is mostly responsible for the retention increase as it increases k*. Furthermore, eq 4 predicts that shallower gradients (e.g. longer t_G values) will lead to larger retention increases when anionic additives are used. This was confirmed experimentally by the results in Table 3 for the peptides upon addition of 20 mM NaClO₄ to the eluent. In the limit of an infinite gradient time, the gradient separation is simply an isocratic separation at the initial eluent strength. Thus, the greatest retention increase for basic compounds when using anionic additives will always be obtained in isocratic elution.

Table 3.

Retention time increase of ten peptides upon addition of 20 mM NaClO₄ in gradient elution

Peptide	ΔtRa (min)
	tG = 15 min	tG = 30 min	tG = 60 min
1	0.48	0.77	1.23
2	0.49	0.79	1.24
3	0.61	1.06	1.88
4	1.10	1.97	3.53
5	1.40	2.54	4.62
6	1.06	1.97	3.63
7	1.36	2.58	4.80
8	0.61	1.11	1.98
9	0.69	1.33	2.49
10	1.74	3.28	6.13

^aRetention time increase from 0 mM to 20 mM NaClO₄ on SB-C18 in mobile phases containing 0.1% TFA. Chromatographic conditions are given in Figure 7.

3.6. Adsorption of Perchlorate Ion to the Stationary Phase

Previous studies [4,14] have shown that hydrophobic anions can adsorb to the stationary phase and increase the retention of basic analytes due to dynamic ion-exchange. To verify the adsorption of the anionic additive in this study, we measured the retention factor of nitrate on the Ace column in 0.1% formic acid with 0 and 20 mM NaClO₄ (see Figure 8) using uracil as the dead time marker. Due to its slight hydrophobicity, the retention time of uracil decreased as %ACN was increased (0.546 min in 24% ACN and 0.511 min in 44% ACN). The average retention time of uracil in different eluents was used to calculate the k’ of nitrate. It is clear that within the range of %ACN studied, nitrate was excluded from the stationary phase (i.e. negative k’) in the presence of 20 mM NaClO₄. In contrast, nitrate was not retained (i.e. nearly zero k’) in the absence of 20 mM NaClO₄. This indicates that some perchorate ions were adsorbed onto the stationary phase. In addition, the degree of exclusion got smaller (i.e. less negative k’) in lower %ACN mobile phase. However, the retention of nitrate in the absence of NaClO₄ also increased as the %ACN was decreased with nearly identical slopes. Therefore, this change is mainly due to the hydrophobicity of the nitrate.

Figure 8.

Retention factor of nitrate as a function of %ACN in two different mobile phases. Amount and type of anionic additive in the mobile phase: (●) None; (○) 20 mM NaClO₄. The average retention time of uracil in the six mobile phases was used to calculate the k’ of nitrate. The stationary phase was Ace 5 C18 and all mobile phases contained 0.1% formic acid. Other experimental conditions: 1.00 mL/min, 40 °C, 210 nm.

Due to the polarity of the perchlorate, the adsorption of perchlorate might decrease the hydrophobicity of the stationary phase. To verify this hypothesis, we measured the retention of a homolog series of alkylphenones on the Ace phase in 0.1% formic acid with 0 mM or 20 mM NaClO₄ (see Figure 9). The identical slopes of the plots of ln k’ vs. number of methylene group clearly indicate that the hydrophobicity of the stationary phase was not changed by the addition of 20 mM NaClO₄. Therefore, the hydrophobic retention of the analyte should not change upon addition of anionic additives.

Figure 9.

Plots of lnk’ vs. n_CH2 of alkylphenone homolog series on Ace 5 C18. Amount and type of anionic additive in the mobile phase: (●) None; (○) 20 mM NaClO₄. All mobile phases contained 36% ACN and 0.1% formic acid. Other experimental conditions: 1.00 mL/min, 40 °C, 254 nm.

3.7. Discussion for the Effect of Anionic Additives on Retention of Cationic Analytes

According to our results, addition of anions to the eluent causes a big decrease in the S values of cationic drugs and peptides but only small changes in their ln k’_w values. The fact that extraction of cations out of water into non-polar solvents such as c-hexane is greatly promoted by addition of strong ion pairing anions seems to be at variance with our observations. We have no explanation for this seeming contradiction other than to remark that over the range in eluent studied here the active stationary phase may not be hydrocarbon-like at all but contains a great deal of sorbed acetonitrile and thus what takes place in the bulk extraction system is not a good model of what takes place in the RPLC.

Further we note that under isocratic conditions, solute retention upon addition of the anions increased to a much greater extent in ACN rich eluents compared to that in water rich eluents. In this section, we discuss this observation by looking at the effect of eluent strength on ion pair formation and the dynamic ion exchange processes.

The retention factor in RPLC is related to the free energy of transferring the solute from the mobile phase to the stationary phase:

$$\ln k’=\ln k_w^{’}-S·\varphi =\ln B-\mathrm{\Delta }{G}_{m\to s}^{\mathit{o}}/\mathit{RT}$$ (5)

where B is the phase ratio, R is the universal gas constant, T is the temperature in Kelvin, m and s denote the mobile and stationary phases respectively. When pure water is used as the mobile phase (i.e. ϕ = 0), eq 5 reduces to:

$$\ln k_w^{’}=\ln B-\mathrm{\Delta }{G}_{w\to s}^o/\mathit{RT}$$ (6)

where w represents the pure water phase. Clearly, ln k’_w values are related to the free energy of transferring the solute from the pure aqueous phase to the stationary phase. When pure organic is used as the mobile phase (i.e., ϕ = 1), eq 5 reduces to:

$$\ln k_w^{’}-S=\ln B-\mathrm{\Delta }{G}_{o\to s}^o/\mathit{RT}$$ (7)

where o represents the pure organic phase. Subtracting eq 7 from eq 6 gives:

$$S=-\mathrm{\Delta }{G}_{w\to o}^o/\mathit{RT}$$ (8)

Therefore, S values are related to the free energy of transferring the solute from the pure aqueous phase to the pure organic phase.

According to eq 8, the magnitude of the S value is determined by the chemical potentials of the analyte (A⁺) and the anion (X^-) in both the aqueous and the organic phases. This equilibrium process is obviously dependent on solvation processes in the water and organic modifier. First, the formation of the ion pair (A⁺X^-) is favored in an eluent with lower dielectric constant (i.e. higher %ACN). This could explain why a larger retention increase was observed in an ACN richer eluent since ion pair formation is known to enhance the retention [4]. In fact, Roses measured the pK_a values of several acids (e.g. acetic acid) in different water-acetonitrile mixtures and found that the pK_a values increased up to 17.5 pH units upon increasing the %ACN from 0 to 100% [36]. This is presumably due to the much stronger association of the acetate and proton in the organic rich eluents. We assume that similar effects will be observed for ion pairing association constants. Clearly this hypothesis needs to be examined experimentally. Second, dynamic ion exchange interactions between adsorbed anions (X^-) and the analyte (A⁺) might also become stronger as the dielectric constant of the mobile phase decreases. Overall, the higher is the %ACN in the mobile phase, the larger is the expected retention increase. In the extreme case of pure water as the mobile phase, both ion pair formation and dynamic ion exchange are strongly disfavored. Thus the addition of the anionic additive might have only a minor effect on the retention of the basic drugs in water rich eluents.

4. CONCLUSIONS

We studied the retention behavior of small and large cations (drugs and peptides) as a function of eluent strength in acidic water-acetonitrile mobile phases with different types and concentrations of anionic additives and different stationary phases. The retention increases upon addition of anions depended strongly on the eluent strength. Upon addition of an anion to the eluent organic rich eluents give larger increases in retention compared to the water rich eluents. It is possible that at very low %ACN mobile phases, addition of anionic additive might even decrease the retention of basic analytes; however, we have not experimentally confirmed this effect. We observed the same behavior on two different stationary phases and in mobile phases with different acidic buffers.

Regression of the retention against the %ACN indicates that the addition of anionic additive substantially decreases S values while only slightly changing ln k’_w values. We believe the decrease in S is the major factor that causes the retention increase of drugs and peptides upon addition of the ion pair former. Therefore, one needs to be cautious when attempting to predict the retention times in the presence of anionic additives in gradient elution by using ln k’_w and S values measured in the absence of anionic additives. We believe this phenomenon is caused by the decreased dielectric constant of the mobile phase as %ACN is increased, which should enhance both ion pair formation in the eluent and dynamic ion-exchange contributions to retention.