Characterizing HPLC Stationary Phases Chromatographic methods can Katarzyna
characterize stationary
Krupczyn´ska and
phases without
Boguslaw Buszewski
destroying them.
Nicholaus Copernicus University (Poland) Pavel Jandera University of Pardubice (Czech Republic)
A
great variety of stationary phases is available to analyze an even larger array of compounds. Stationary phases in which organic moieties (alkyl chains, aryl, amine, cholesterolic, phospholipids, phenol groups, etc.) are grafted onto a silica surface are the most popular. Sometimes the stationary phases themselves must be characterized, such as for quality control purposes. Often, interpreting chromatographic data is difficult because even columns that are “identical”—produced by the same manufacturer—exhibit different chromatographic properties. Determining the structure and properties of new stationary phases is also extremely important so that the retention mechanism, which plays a significant role in selectivity, can be described accurately. Chromatographic separation occurs because of the differences in the affinities of solute molecules for the stationary phase—retention is a process of solute transfer from a mobile phase onto/into a stationary phase (1). Column quality depends on a number of factors, the most important of which are homogeneity of both particle packing in the column bed and arrangement of the chemically bonded phase on the adsorbent surface. These factors determine the reproducibility of chromatographic data, and the large number of these factors is why there are a lot of methods for characterizing columns. Obviously, no ideal single technique can characterize all the
© 2004 AMERICAN CHEMICAL SOCIETY
J U LY 1 , 2 0 0 4 / A N A LY T I C A L C H E M I S T R Y
227 A
properties of various packings. Many physicochemical techniques are used for this purpose: FTIR; cross-polarization magic-angle spinning (MAS) NMR for 29Si and 13C; electron, atomic, and tunneling microscopies; and thermochemical and thermogravimetric methods. However, these techniques damage columns during analysis. To avoid this, chromatographic tests that evaluate quality and analytical suitability on the basis of intermolecular interactions between the analyte, the stationary phase, and the mobile phase can be applied.
Pores and surface area The physical sorption of gases and vapors on porous solids is a very popular and powerful tool for determining surface area, porosity, and pore-size distribution and evaluating kinetic and thermodynamic values (2). These factors have a significant influence on the synthesis of the phase, calculation of ligand density, column efficiency, and selectivity. Gas uptake is measured as a function of the partial pressure of the gas at equilibrium (3). The characteristic shape of the nitrogen adsorption isotherm is sigmoidal and has three distinguishable regions: the concave region at low pressures, a linear region for partial pressure, and a high-pressure convex region (2). The results are plotted as p/[V(p0 – p)] versus the relative pressure ratio p/p0, in which p is the gas pressure, p0 is the saturation gas pressure at a given temperature, and V is the volume of gas. An alternative method for determining the pore-size distribution would be to measure the molecular weight distribution by gel permeation chromatography using a series of polystyrene standards (4). This technique could also be used for characterizing the support surface before and after chemical modification. Organic ligands can block adsorbent pores so that the retention volume VR of non-retained solutes decreases for chemically bonded stationary phases. Gebet and Kováts proposed chemisorption of trimethylsilyl or (3,3-dimethylbutyl)dimethylsilyl groups as another method for determining surface area (5). First, the amount of space needed for the ligands was determined, and then the surface was modified with a silane under controlled conditions. The surface area was calculated using the phase loading values and the appropriate space requirement values. Pore accessibility can be determined using small-angle neutron scattering in which low-energy neutrons strike the sample and are elastically scattered by pores or other structures (6). Detailed information can be obtained about the size and surface area of the probed scattering pattern within ~1–100 nm (2). Microscopic measurements provide very interesting information about the pore structure of packing materials (3, 7 ). Tanaka et al. investigated the internal structure of glass, silica particles, and polymer gels by transmission electron microscopy (TEM; 8). Nitrogen adsorption and size-exclusion chromatography only provide information on overall porosity and poresize distribution of particles, but TEM complements these 228 A
A N A LY T I C A L C H E M I S T R Y / J U LY 1 , 2 0 0 4
techniques by providing useful information about the shape, location, and uniformity of pores.
Elemental, thermal, and microcalorimetric analysis Elemental analysis is one of the basic methods for evaluating stationary phases (2, 9). The quantity of carbon, nitrogen, and hydrogen can be measured directly by burning the sample in oxygen. Knowing the percentage of carbon allows the coverage density to be calculated according to the Berendsen equation (10) RIP =
106 PC 1200nC – PC(M1 – nx)
·
1
(1)
S BET
in which RIP is the surface coverage density, PC is the percentage of carbon, nC is the number of carbon atoms per silane moiety, M1 is the molar weight of silane, nx is the number of reactive groups in the silane, and SBET is the specific surface area of unmodified support. To endcap or prepare “pseudobilayer” phases, we modified this equation to RIIP =
PC(2) + PC(2)RIIP(M1–1) – 1200RIP nC(1) 1200nC(2) – PC(2)(M2 – 1)
·
1 S BET
(2)
in which RIIP is the surface coverage density, PC(2) is the percentage of carbon bonded in the endcapping step, nC(1) is the number of carbon atoms in the first stage, nC(2) is the number of carbon atoms in the second stage, M2 is the molar mass of the modifier in the endcapping process, and SBET is the specific surface area of unmodified support (9). Unfortunately, elemental analysis does not provide information about the homogeneity of the stationary phase and should be used only as a complement to other more sophisticated analytical characterization methods. Thermal analysis is a useful technique for the determination of organic ligands on the modified adsorbent surface (mainly silica gel). Thermal gravimetric analysis is one of the simplest methods for determining surface characteristics. The weight loss after burning the sample in oxygen is equal to that of the chemically bonded phase, provided that the loss of other moieties, such as water, is avoided. The results from differential thermal analysis enable calculation of the surface area ratio of bare silica to carbon-coated silica (2). Differential scanning calorimetry (DSC) is also a valuable tool for characterizing chemically modified surfaces. Hansen and Callis investigated phase and structural transitions of an adsorbed film on a microporous solid surface as methanol was added (11). Investigating the parameters characterizing the phase transition helped explain the association and dissociation of methanol with the chemically bonded chains. The shape of the DSC curve depends on the surface porosity and provides information on the structural heterogeneity of the stationary phase (9). Enthalpy effects related to the temperature changes in water/silica gel systems can also be studied. All these experiments confirm Serpinet’s observations regarding the shape and transition of chemically bonded ligands (12, 13). Adiabatic calorimetry may also be used
26
Intermediate
24 22 Bare silica gel
20 18
% Transmittance
16 14
Final
12 10 8 6 4 2 0 4000
3600
3200
2800
2400 Frequency
2000
1800
1600
1400
(cm–1)
FIGURE 1. FTIR spectra of bare silica gel and the intermediate and final products of a cholesterolic phase. (Adapted with permission from Ref. 23.)
to estimate transition entropy and enthalpy and provide confirmation of chemically bonded moieties (14). The presence of a chemically bonded phase changes the physicochemical or chromatographic properties of the support surface, such as the free surface energy. The free surface energy of the packing significantly influences the nature and magnitude of adsorption, which can be characterized by the wetting heat. On the molecular level, wetting is influenced by chain–chain, chain–eluent, and residual silanols–eluent interactions, and some energy is evolved during these interactions. We measured the heats of wetting of various solvents using microcalorimetry (9, 15). Knowing the changes in the phase and the heats of wetting makes it possible to characterize the surface of the silica before and after modification. Surface wettability helps to define the conformation of chemically bonded phases and to predict solvent penetration into the stationary phase, because it depends on the organization of the brush-type ligands.
Spectroscopic methods Fluorescence. Hunnicutt and Harris used X-ray fluorescence for depth analysis and X-ray photoelectron spectroscopy for surface analysis to characterize adsorbents (16). Comparing the results of the two techniques gives a picture of the differences between the inner pore network and the outer particle surface. Lochmüller suggested that three types of fluorescent ligands chemically bond to the silica support: those located inside micropores or surrounding the alkyl chains where the solvent molecules cannot penetrate between moieties, those “open” for solvent molecules, and those relatively accessible for solvent molecules but bonded with silanols present on the surface (17). The type
of ligand and the composition of the mobile phase control the polarity of the stationary phase. Several researchers adapted the fluorescence characteristic of pyrene (chemically bound on the surface of the silica support and adsorbed) to investigate the polarity of alkyl-modified surfaces of reversed-phase packings and how polarity depends on the composition of the surrounding mobile phase (18). The time-dependent luminescence of the variable surface concentration of bound pyrene on silica can be used to investigate the distribution of the bound molecules and their organization in contact with different solvents. Bogar used the formation of pyrene excimers to study the lateral diffusion of solutes in a C18 bonded phase (19). IR. In early applications, the main goal of this technique was to prove the existence of bonded ligands on the support surface. Berendsen and de Galan assigned characteristic absorption bands to the specific ligand functionalities (20). As the methods were developed and improved, the kinetics of modification and dynamics of bonded ligands were investigated (21, 22). FTIR with signal averaging overcame the problems connected with high absorption of IR radiation by silica gel (2). Note that the particle diameter of the chromatographic substrate is similar to the wavelength of IR radiation, and the signal of interest can be superimposed on a background signal. This problem can be eliminated by careful subtraction and baseline linearization, as suggested by Sander et al., who compared the spectra of C1–C22 bonded phases with those of the corresponding chlorosilanes (22). They observed that the alltrans form is the most ordered state for alkyl chain conformation. The end gauche conformer is a little less ordered, but it gives rise to a transition band at = 1341 cm–1. The gauche– J U LY 1 , 2 0 0 4 / A N A LY T I C A L C H E M I S T R Y
229 A
2´, 3´, 1, 8 – 27 21
CH3 22 23 27 CH3 CH 24 25 CH3 20 19 12 17 CH3 16 CH3 11 13 14 15 26 18
H O
2 3
Si – CH2 – CH2 – CH2 – N – C – O 3´
4´
1 4
10 5
9
8 6 7
1´
2´
4´ 2,4 7
5
1´
150
(ppm)
3
6
100
50
0
FIGURE 2. 13C CP/MAS NMR spectrum of cholesterolic packing.
gauche conformation with a nearly 90° bend in the alkyl chain provides the absorption band at = 1350 cm–1. The kink conformer ( = 1367 cm–1) has parallel but laterally displaced segments. In addition to alkyl-bonded phases, FTIR can be used to characterize phases containing other ligands (cholesterolic, alkylamide, etc.; Figure 1; 23, 24). Photoacoustic. Low and Parodi used photoacoustic spectroscopy (PAS) for the first time as a complementary technique to FTIR to investigate highly light-scattered samples of bare and modified silica (25). PAS is suitable for measuring high-resolution spectra of solid–liquid slurries, which greatly facilitates studying the bonded phases in a mobile-like phase (2).
NMR. In the beginning, there were problems with line broadening caused by the mobility of bonded ligands, chemical shift anisotropy, carbon–proton dipolar interactions, and long spinlattice relaxation times. MAS at rates >10,000 Hz, cross-polarization (CP), high-power enhanced sensitivity, and resolution of solid-state measurements overcame these difficulties (26). Thanks to CP, proton magnetization is transferred during a distinct contact time to diluted heteronuclei, such as 29Si or 13C. The presence of a rigid system and attached protons leads to an effective magnetization transfer and high signal intensities. At a longer contact time, more flexible moieties of the stationary phase are polarized, and variations of contact time can be used to investi-
Q3 Q4 Q2
CH3 O Si CH2 R
OH
OH Si
O Si CH2 R
CH3
OH
M
T1
OH
M
Q2 D1
CH3 O Si CH2 R OH D1
O CH3 Si O CH2 R D2
Si OH Si OH OH O Si CH2 R O O Si CH2 R O O Si CH2 R OH
T2 T3 T2
D1
D2 D2
T2
Q3
T3
T1 T2
O O Si O O Q4
T3
T1
(ppm) 0
–20
–40
–60
–80 –100 –120
FIGURE 3. Structures of chemically bonded ligands on a silica structure obtained by 29Si CP/MAS NMR. (Adapted from Ref. 28.) 230 A
A N A LY T I C A L C H E M I S T R Y / J U LY 1 , 2 0 0 4
Trans Gauche
as the plate number, and silanol activity (31). The elution se60,000 quence of neutral, polar, and 40,000 basic benzene derivatives can 20,000 0 be used to describe reversed– 20,000 phase behavior (32). The au– 40,000 (1H) (Hz) thors of this test focused on – 60,000 – 80,000 characterizing hydrophobicity, 0 50 25 silanol activity, and shape se (13C) (ppm) lectivity and established that the retention of neutral solutes depends on the carbon loading, silica porosity, and lengths FIGURE 4. Dynamic behavior of C30 ligands characterized by 2-D solid-state NMR. of chemically bonded organic (Adapted with permission from Ref. 52.) ligands. The stationary phase, where aniline is eluted before gate the dynamic properties of adsorbents (27 ). 13C CP/MAS phenol and the three isomers of toluidine are not separated, NMR also provides useful information about the surface struc- suppressed silanol activity and is suitable for separating basic compounds. ture of chemically bonded ligands (Figure 2; 23). Tanaka et al. determined that the number of alkyl chains In some of the first work in this area, Maciel and Sindorf used 29 Si CP/MAS NMR to obtain the signals characteristic for various (A), hydrophobicity (B), steric selectivity (C), hydrogen bondfeatures of the silica structure (28). Isolated silanols were identified ing capacity (D), ion exchange capacity at pH >7 (E), and ion at = –99.8 ppm, siloxane silanols at = –109.3 ppm, and gemi- exchange capacity at pH
