3 STATIONARY PHASES Yuri Kazakevich and Rosario Lobrutto 3.1 INTRODUCTION The column is the only device in the high-performance liquid chromatography (HPLC) system which actually separates an injected mixture Column packing materials are the “media” producing the separation, and properties of this media are of primary importance for successful separations Several thousands of different columns are commercially available, and when selecting a column for a particular separation the chromatographer should be able to decide whether a packed, capillary, or monolithic column is needed and what the desired characteristics of the base material, bonded phase, and bonding density of selected column is needed Commercial columns of the same general type (e.g., C18) could differ widely in their separation power among different suppliers Basic information regarding the specific column provided by the manufacturer, such as surface area, % carbon, and type of bonded phase, usually does not allow prediction of the separation or for the proper selection of columns with similar separation patterns Great varieties of different columns are currently available on the market Four distinct characteristics could be used for column classification: Type (monolithic; porous; nonporous) Geometry (surface area; pore volume; pore diameter; particle size and shape; etc.) HPLC for Pharmaceutical Scientists, Edited by Yuri Kazakevich and Rosario LoBrutto Copyright © 2007 by John Wiley & Sons, Inc 75 76 STATIONARY PHASES Surface chemistry (type of bonded ligands; bonding density; etc.) Type of base material (silica; polymeric; zirconia; etc) All these characteristics are interrelated Variations of porosity which include pore diameter can affect both the adsorbent surface area and the bonding density The type of base material affects adsorbent surface chemistry Therefore, in our discussion we combine these characteristics in two major classes: geometry and surface chemistry Most geometry-related properties of packing materials are related to the column efficiency and flow resistance: particle size, particle shape, particle size distribution, packing density, and packing uniformity Surface-chemistryrelated properties are mainly responsible for the analyte retention and separation selectivity Adsorbent surface area, pore volume, and pore diameter are the properties of significant importance HPLC retention is generally proportional to the surface area accessible for a given analyte (Chapter 2) Surface area accessibility is dependent on the analyte molecular size, adsorbent pore diameter, and pore size distribution The chemical nature of the ligands bonded on the surface of support material defines the main type of chemical interactions of the surface with eluent and analyte molecules In essence, all C18-type columns should be similar with regard to their main interaction type, namely, hydrophobic interactions: Methylene selectivity of all C18-type columns are virtually identical [1] Bonded phases of the same type differ in their ability to suppress (or shield) other types of interactions (ionic; dipole) exerted by the base material (e.g., silica) Energy of these unwanted interactions is about 10 times greater than the energy of dispersive interactions [2] Due to the exponential nature of the relationship between retention and interaction energy even the presence of 1% or less of these active centers in the packing material surface can significantly affect the analyte retention Bonding density is the primary parameter in evaluation of the quality of the bonded material Usually the higher the bonding density, the better the shielding effect, although care should be taken in cross-evaluation of similar columns on the basis of their bonding density Surface geometry can also significantly affect bonding density Base material with smaller pores has higher surface area; however, bonding density is usually lower due to the smaller pores All parameters of the packing material are interrelated in their influence on the chromatographic performance of the column The quality of an HPLC column is a subjective factor, which is dependent on the types of analytes and even on the chromatographic conditions used for the evaluation of the overall quality Long-term column stability (pH and temperature) and batch-to-batch reproducibility are probably the most important quality characteristics to be considered in column selection in the pharmaceutical industry Nevertheless, BASE MATERIAL (SILICA, ZIRCONIA, ALUMINA, POLYMERS) 77 these criteria should be evaluated with caution when selecting the column evaluation parameters Long-term stability of retention and efficiency characteristics are usually different, depending on the testing conditions (mobile phase, temperature, and analyte probes) Efficiency is usually fairly stable at low mobile-phase pH, while retention of the probe analytes may show a drift in retention However, the retention is generally stable at high pH while efficiency could be deteriorated 3.2 TYPE OF PACKING MATERIAL (POROUS, NONPOROUS, MONOLITHIC) Majority of packing materials used in HPLC are porous particles with average diameters between and 10 µm For most pharmaceutical applications, 3-µm particle sizes are recommended Porosity provides the surface area necessary for the analyte retention (usually between 100 and 400 m2/g) Interparticle space is large enough to allow up to 1–3 mL/min flow within acceptable pressure range (however, the pressure drop across the column depends on the particle size, length of column, temperature of separation, and type of mobilephase composition) Introduction of small nonporous spherical particles in the mid-1990s [3, 4] was an attempt to increase efficiency by eliminating dual column porosity In the column packed with porous particles, interparticle space is about 100-fold larger than pores inside the particles, and liquid flow around the particles is also faster; this leads to the significant band broadening Unfortunately, elimination of particle porosity dramatically decreases adsorbent surface area, thereby decreasing the column loading capacity Columns packed with small (1.5 µm) nonporous particles also require ultra-microinjection volumes and a corresponding increase of detector sensitivity The introduction of monolithic columns in the 1990s was another and more successful attempt to increase column permeability while decreasing the gap in column dual porosity Macropores in the monolith are between 4000 and 6000 Å in diameter, and they occupy almost 80% of the column volume Compared to the conventional packed column with 5- or even 3-µm particles, the silica skeleton in monolith is only approximately µm thick, which facilitates accessibility of the adsorbent surface inside the mesopores of the skeleton (pores between 20 and 500 Å in diameter are usually called mesopores) Comparison of the spherical packing material and monolithic silica is shown in Figure 3-1 3.3 BASE MATERIAL (SILICA, ZIRCONIA, ALUMINA, POLYMERS) In modern liquid chromatography, almost all reversed-phase separations are performed on chemically modified adsorbents Analyte interactions with the 78 STATIONARY PHASES Figure 3-1 SEM pictures of HPLC silica particles (5 µm) and silica monolith (Reprinted with permission from reference 5.) stationary phase surface are the primary factor for successful separations Most commercial adsorbents reflect their surface chemistry in their names (e.g., C18, C8, Phenyl, etc.) while the base material used usually is not specified, although its properties are very important Specific parameters of the base of packing material are: • • • • • • • • • • • Surface area Pore size Pore volume Pore size distribution Particle shape Particle size Particle size distribution Structural rigidity Chemical stability Surface reactivity Density and distribution of the surface reactive centers Surface area is directly related to the analyte retention [Equation (2-47) in Chapter 2] Generally, the higher the surface area, the greater the retention Pore size is a critical parameter for the surface accessibility Molecules of different size could have different accessible surface area due to the steric hindrance effect (bigger molecules might not be able to penetrate into all pores) Pore size is also related to the surface area Assuming that all pores of the base material are cylindrical and neglecting the networked porous structure (assuming straight and not interconnected pores), it is possible to write the following expressions for the surface area and pore volume [6]: S = 2pRL, V = pR L (3-1) where S is a surface area of one gram of porous adsorbent; R is the average pore radius; V is a pore volume of one gram of adsorbent; and L is a total length of all pores in the same one gram of the adsorbent 79 BASE MATERIAL (SILICA, ZIRCONIA, ALUMINA, POLYMERS) It may be interesting for the reader to estimate an approximate length of all pores in g of average adsorbent Surface area of average HPLC adsorbent is on the level of 300 m2/g and average pore diameter is 100 Å One gram of silica is an approximate amount that is usually packed into the standard 15-cm-long HPLC column (4.6-mm I.D.) If you calculate the length of all pores in this column [using equation (3-1) and express it in meters or kilometers], you will get a feeling of what you are dealing with when you are using HPLC If we take a ratio of the above expressions, we get simple relationship between these parameters: S = V R (3-2) K K Unger [6] found that in most cases, expression (3-2) shows a 15% discrepancy between measured and estimated adsorbent surface, which is very good when we take into account the above assumptions made in its derivation The most commonly used base material is silica (SiO2), the most common substance on the Earth and thoroughly studied in the last two centuries An excellent monograph on the properties of silica was published by Iler [7] Development of modern HPLC techniques promoted advancement in porous silica technology Almost all silica-based HPLC packings manufactured in the twenty-first century are very uniform spherical porous particles with narrow particle and pore size distribution Silica has one significant drawback: It is soluble at high pH, although chemical modification with high bonding density of attached alkylsilanes extends its stability range to over pH 10 Another porous base material suggested in the last decade as an alternative to silica is zirconia Zirconia is stable in a very wide pH range (pH 1–14), but zirconia surface has relatively low reactivity (more difficult to bond different functional groups to the surface), which significantly limits a selection of available stationary phases Polymer-based materials have been on the market for more than 30 years Crosslinked styrene-divinylbenzene and methylmethacrylate copolymers are the most widely used These materials show high pH stability and chemical inertness Their rigidity and resistance to the swelling in different mobile phases is dependent on the degree of crosslinkage Practical application of these materials for the separation of small molecules are somewhat limited due to the presence of microporosity Gaps between cross-linked polymer chains are on the level of molecular size of lowmolecular-weight analytes These analytes could diffuse inside the body of a polymer-based packing material, which produce drastically different retention of a small portion of injected sample than the rest of it At the same time, polymers are the main packing material for size-exclusion chromatography 80 STATIONARY PHASES 3.4 GEOMETRY 3.4.1 Shape (Spherical/Irregular) Recent technological advancements made spherical particles widely available and relatively inexpensive Columns packed with spherical particles exhibit significantly higher efficiency, and columns packed with irregular particles are seldom used and are becoming nonexistent for analytical scale separations 3.4.2 Particle Size Distribution Packing materials are characterized by the average diameter of their particles and the distribution of the particle size around the average value The particle size distributions shown in Figure 3-2 are for spherical packing materials with nominal particle size of 10 µm Distributions of different batches are symmetrical, with average width of approximately 50% of nominal diameter Most critical for HPLC application is the presence of very small particles (fines) less than 0.5 µm These small particles are usually fragments of crushed particle (porous silica is a fragile material) These fine particles will steadily migrate in the column toward the exit frit and clog it These particles will eventually dramatically decrease the column efficiency, and peak distortion is usually observed for all peaks in the chromatogram Particle size distribution itself does not affect chemical behavior of HPLC adsorbent, although it is known to influence the efficiency of packed column Packings with wide particle size distribution contain a significant amount of Figure 3-2 Example of cumulative particle size distribution of HPLC packing materials (Waters µ-Bondapack) Average particle size is 10 µm (inflection point) (Reprinted from reference 8, with permission.) 81 GEOMETRY fine particles, which increases column backpressure; a big size difference between particles in the column decreases overall column efficiency Halasz and Naefe [9] and Majors [10] suggested that if the distribution is not wider than 40% of the mean, then acceptable flow resistance and column efficiency can be obtained The narrower the particle size distribution, the better and the more reproducibly the columns could be packed Generally accepted criteria is that 95% of all particles should be within 25% region around the mean particle diameter [11, 12] 3.4.3 Surface Area Surface area of HPLC adsorbents is probably the most important parameter, although it is almost never used or accounted for in everyday practical chromatographic work As shown in the theory chapter (see Chapter 2), HPLC retention is proportional to the adsorbent surface area The higher the surface area, the greater the analyte retention, although as we discuss later, depending on the surface geometry, analytes of a different molecular size could effectively see different surface areas on the same adsorbent The experimental methods for the measurement of the surface area of porous silica is fairly well established Nitrogen or argon adsorption isotherms at the temperature of liquid nitrogen (77 K) are used in accordance with BET (Brunauer, Emmet, Teller) theory [13] for the calculation of the total surface area per unit of adsorbent weight.There are different variations of BET theory available as well as different instrumental approaches [14] for the measurement of nitrogen isotherms For proper characterization of mesoporous (pore diameter is greater than 20 Å) adsorbents, the static measurement of adsorption isotherm with proper equilibration at each measured point is preferable Detailed discussion of all aspects of nitrogen adsorption isotherms and related theories could be found in the classic book Adsorption, Surface Area and Porosity by Gregg and Sing [15] Full nitrogen adsorption isotherm (adsorption and desorption branches) is shown in Figure 3-3 The region between 0.05 and 0.25 relative pressures is called the BET region, and it is used for the determination of the so-called monolayer capacity—the amount of nitrogen molecules adsorbed on the sample surface in a compact monolayer fashion The BET equation represents the dependence of amount of adsorbed nitrogen as a function of the relative equilibrium pressure (p/p0): p p0 p n − p0 = (C − 1) p + nm C nm C p0 (3-3) where nm is the monolayer capacity, C is energetic constant of nitrogen interaction with the surface; p/p0 is the relative equilibrium pressure, and n is the 82 STATIONARY PHASES Figure 3-3 Nitrogen adsorption isotherm 1, Adsorption branch; 2, desorption branch amount adsorbed Figure 3-3 shows the experimental dependence of the amount of nitrogen adsorbed on the surface versus the relative pressure (pressure of nitrogen at the equilibrium in the gas phase over the adsorbent related to the saturation pressure at the temperature of the experiment) Equation (33) is the linear form of the function shown in Figure 3-3, but only in the relatively low pressure region between 0.05 and 0.25 where the formation of adsorbed monolayer is complete and BET theory is valid The plot of the experimental points in p/p0/(1 − p/p0) versus p/p0 allows for linear minimization and calculation of C and nm values [15] It is generally assumed that a nitrogen molecule occupies 16.4 Å2 on the polar silica surface The adsorbent surface area is then calculated as a product of the total amount of nitrogen in the monolayer (nm) and the nitrogen molecular area (16.4 Å2) 3.4.4 Pore Volume At higher relative pressures, above 0.7 in Figure 3-3 a fast increase of the adsorbed amount of nitrogen is observed This region is attributed to the process of capillary condensation of nitrogen inside the adsorbent pores This increase is observed until the whole pore volume is filled with liquid nitrogen When relative equilibrium pressure approaches the saturation pressure and all pores are already filled with liquid nitrogen, a small flat section on the adsorption isotherm is usually observed (amax) This section indicates the completion of the pore filling with condensed nitrogen, and it could be used for accurate determination of the adsorbent pore volume: 83 GEOMETRY Vpore = VL amax P RT (3-4) where VL is the molar volume of liquid nitrogen (34.7 mL/mol); amax is the maximum amount of nitrogen in the pores expressed in milliliters at atm and L ⋅ atm 25°C; P is the pressure (1 atm); R is the gas constant ( 0.082 ); and T is K ⋅ mol the temperature, 298 K The desorption branch of nitrogen isotherm is typically used for the determination of the pore size distribution.The only important factor that should be carefully verified for each adsorbent is the presence of microporous structure If the micropores (pores with diameter less than 20 Å) are present in base material, the actual surface to which HPLC analyte might be exposed will be different from the surface measured with nitrogen adsorption This is due to the size difference of nitrogen molecule and practically any HPLC analyte molecule Bigger molecules will have steric hindrance in micropores, and any interpretation of the HPLC retention related to the surface area will be erroneous In addition, proper chemical modification of adsorbents with micropores is essentially not possible Minimum pore diameter acceptable in HPLC adsorbents is approximately 50 Å Adsorbent pore size provided by the manufacturer is the diameter corresponding to the maximum of the pore size distribution curve, obtained from the adsorption branch of nitrogen isotherm The distribution of the pores could vary significantly as it is shown in Figure 3-4 Figure 3-4 Pore size distribution of different HPLC materials Allure Silica (Restek); Allure-PFPP (Restek), Prodigy-Silica (Phenomenex); Chromolith C18 (Merck KgaA, Germany); and research-type ordered silica with highly uniform pores of 50-Å pore diameter 84 STATIONARY PHASES 3.4.5 Surface Geometry The roughness of the silica surface could introduce the steric hindrance of the surface accessibility similar to the effect of the micropores In the discussion above, we assume the ideal tubular geometry of the silica surface The use of different probe molecules for the BET measurement of silica surface area (such as N2, Ar, Kr, benzene, etc.) leads to significant difference in the surface area values for the same silica sample It was suggested that silica surfaces possess the property of fractals [16]; this essentially means that molecules of different size will see a different surface area The surface constructed of ridges and valleys could be considered as an example of a fractal surface The slopes of these ridges are also constructed by smaller ridges, and with the higher magnification even smaller ridges are visible As an example, if a big ball is used to roll over this surface, it will see only the big ridges and thus a relatively small total surface If, on the other hand, a smaller ball is rolling over the same surface, it will see much more of smaller ridges and a lot of them, resulting in a much higher effective surface area The smaller the probing ball, the finer the surface roughness it will see and correspondingly higher surface will be detected Molecular nitrogen will see a significant surface area due to its small size comparable to the dimensions of the surface roughness, while bigger molecules such as pyrene will not be able to see all ridges and valleys and will see a significantly lower effective surface area These factors have been studied extensively [17] for silica, and authors have found fractal factors to vary between and 3, depending on the silica synthesis, treatment, and so on Adsorbent surface area (S) is measured as a product of molecular area (s) of a probe substance and the number of the molecules (N) in complete adsorbed monolayer On the fractal surface the total number of molecules in the monolayer is dependent on its roughness and could be expressed as N ~σ − D (3-5) where D is a fractal number Since S = Nσ, the adsorbent surface could be expressed as follows: S ~ s ⋅ N ~ s ( 2−D ) (3-6) On the flat surface, D is equal to and only in this case the surface area is not dependent on the size of probe molecule The higher the fractal number, the less accessible the surface (quasi-threedimensional or rough surface) For silica with a pore diameter of 10 nm and higher, the fractal factor has a tendency to be between 2.05 and 2.3, which is close to the flat surface Figure 3-5 illustrates the apparent decrease of accessible silica surface in the form of a fraction of the total surface with the increase of the fractal number of this surface (roughness) For these types of ... in HPLC are porous particles with average diameters between and 10 µm For most pharmaceutical applications, 3-µm particle sizes are recommended Porosity provides the surface area necessary for. .. compounds) could play a major role All these factors are the driving force for the search in alternative base materials for HPLC packings 90 STATIONARY PHASES The majority of polymer-based packing... predictability of retention and specificity for the separation of conformational isomers, and similar advantages are expected for these adsorbents in HPLC SURFACE OF CHEMICALLY MODIFIED MATERIAL