Today, organometallic complexes are of great interest because of their applications in bond activation processes and catalysis. In the field of chromatography, organometallic complexes are used for the separation of unsaturated organic compounds based on their interactions with metals immobilized in the stationary phase (Guha and Janak, 1972), especially silver-ion chromatography uses the capability of unsaturated compounds to form organometallic complexes containing silver(I) ions (Mander and Williams, 2016; Williams and Mander, 2001). Unsaturated compounds form weak reversible com- plexes of different strengths with silver ions immobilized in the stationary phase, during their elution through the chromatographic column. It is a dynamic system with continuous establishment of equilibrium among complexed and free components with a high equilibrium constant. Complexes are of the charge-transfer type, where unsaturated compounds donate electrons to the silver ion (acceptor).
The description of complexation bonding between DB and silver(I) ion by the DewareChatte Duncanson model (Dewar, 1951; Chatt and Duncanson, 1953) is now widely accepted. This model describes the stabilization of complexes as a combination of s-donation and p-back-bonding in- teractions between DB and the metal, i.e., donation ofp-electrons from the occupied 2p bonding orbital of the olefinic DB into vacant 5sand 5porbitals of the silver ion (s-type bond,Fig. 4.1A), and the back-donation ofd-electrons from occupied4dorbitals of the silver ion into unoccupied p-2p antibonding orbitals of the olefinic DB (p-type bond,Fig. 4.1B). This model describes only bonding between DBs and silver ions, but the stability constant of the complex also depends on steric and polar effects. A number of experimental and theoretical studies have been done for complexes of the silver ion with short olefins because these complexes are important in organometallic chemistry. Early studies of stability constants of organosilver complexes used distribution methods based on the distribution of organic compounds between an organic phase and silver nitrate aqueous solution (Lucas et al., 1943; Winstein and Lucas, 1938). These results have been confirmed later by various analytical techniques, i.e., UV, infrared, and Raman spectroscopy, based on shifts in absorption maxima between complexed and free unsaturated compounds (Hosoya and Nagakura, 1964), X-ray studies of organosilver monocrystals (Gmelin, 1975; Bressan et al., 1967), electron spin resonance (Kasai et al., 1980), etc. General conclusions of complex stability affected by structural factors are as follows: The stability of complexes containingcis-DB is higher than withtrans-DB (Lucas et al., 1943; Morris, 1966); complexes of methylene-interrupted DB are stronger than conjugated ones; the stability of
FIGURE 4.1
Description of complex bonding between silver ions and double bonds (DBs) by the DewareChatteDuncanson model: (A)s-donation and (B)p-back-bonding interactions between the metal and DB.Ag, silver ion.
Redrawn with permission from Dewar, J.S., 1951. A review of the pi-complex theory. Bulletin De La Societe Chimique De France 18, C71eC79.
complexes increases with increasing distance of DB (Winstein and Lucas, 1938); the stability of complexes decreases with increasing chain length (Conacher, 1976); and the stability of complexes increases by substitution of hydrogen with deuterium atoms (Cvetanovic et al., 1965).
In the silver-ion chromatographic process, interactions between DBs of unsaturated compounds and silver ions are rather complex. Electron spin resonance shows the interaction of a silver ion with two molecules (Kasai et al., 1980), and X-ray diffraction of a monocrystal shows the coordination of one silver ion with two DBs from different molecules (Gmelin, 1975; Bressan et al., 1967). The interaction of a silver ion with the carboxylic oxygen of unsaturated compounds has also been shown (Winstein and Lucas, 1938). In addition to many interactions of silver ions and unsaturated compounds, the retention is also influenced by the quality of column packing, i.e., the density and accessibility of silver ions on the surface of the stationary phase. Nowadays, most Ag-HPLC columns are based on the silica matrix chemically modified with an alkylphenylsulfonic moiety, with bonding of silver ions by ionic bonds. Free silanol groups may interact with unsaturated molecules during the chromatographic process, and molecules are separated based on mixed retention mechanisms. The separation of lipids is more complex because of a number of combinations of interactions among silver ions, DBs, and carboxyl oxygen atoms. Electrostatic forces within and between fatty acyl chains could also influence their retention behavior, but little is known about these forces so far.
3. PARAMETERS AFFECTING SILVER-ION HIGH-PERFORMANCE LIQUID CHROMATOGRAPHY
3.1 TYPES OF SILVER-ION SYSTEMS
Three potential ways for embedding silver ions in the HPLC system have been described so far (Momchilova and Nikolova-Damyanova, 2003; Nikolova-Damyanova, 2005, 2009): (1) adsorption of silver ions on the stationary phase, (2) silver ions embedded in the stationary phase via ionic bonds, and (3) addition of silver ions into the mobile phase.
1. Silver complexation column (Jeffrey, 1991; Schuyl et al., 1998)din the past, silver-ion columns for HPLC were prepared by the adsorption of silver ions (typically in the form of silver nitrate) on silica columns and then washed carefully with different solvents and the final washing with the mobile phase. A certain part of silver ions is adsorbed on the silica gel. Some papers present good separations with this type of column, but at present they have almost disappeared because of serious drawbacks, such as the leakage of silver ions into the mobile phase [critical for mass spectrometry (MS) coupling], poor reproducibility, and also technical skills required for the reproducible preparation of columns.
2. Strong cation exchanger modified with silver ions (Emken et al., 1964; Christie, 1987; Toschi et al., 1993)dsilver ions replace initial protons in theeSO3H functional groups (estimated silver content is 50e80 mg) and theneSO3Ag interacts with DB during the chromatographic process.
This ionic bond is rather stable and no leakage of silver ions is detected even for long-time use in high-performance liquid chromatographyemass spectrometry (HPLC/MS) experiments.
Nowadays, practically all Ag-HPLC/MS papers are based on the ion-exchanger type of silver-ion columns. The commercial silver-ion columns, Chromspher Lipids, can be now purchased from Agilent Technologies. Some researchers rely on their laboratory-made silver-ion columns with
comparable performance according to the procedure published byChristie (1987).
Recommended eluents for silver-ion columns are dichloromethane, dichloroethane, acetone, acetonitrile, toluene, and ethylacetate. Aqueous solvents are discouraged because they permanently alter column properties, especially small anions can cause silver precipitation.
Solvents should be free of peroxides or any reducing agents that can cause the reduction of silver(I) to the metal state. No acids should be used because of the back replacement of silver ions to protons.
3. The addition of silver ions into the mobile phase using C18 column (Correa et al., 1999;
Nikolova-Damyanova et al., 1993). This approach is not compatible with MS detection because nonvolatile inorganic salts cause contamination of the ion source and because of ion suppression effects. The retention mechanism in this arrangement is a combination of silver ion and reversed- phase (RP) modes.
3.2 MOBILE-PHASE COMPOSITION
Optimization of the mobile phase is a crucial step in Ag-HPLC because the proper optimization of solvent composition and gradient steepness can significantly improve the separation, including the regioisomeric resolution. Two types of mobile phases are most frequently used in Ag-HPLC. The first type is based on chlorinated solvents, such as dichloromethane or dichloroethane with the addition of other polar modifiers at low concentration, typically acetonitrile, acetone, or methanol (Christie, 1988;
Juaneda et al., 1994; Laakso and Voutilainen, 1996; Nikolova-Damyanova et al., 1992, 1995a,b;Lı´sa et al., 2013). The second type is hexane-based mobile phases with the addition of acetonitrile as the polar modifier (Adlof and List, 2004; Adlof, 1994, 1995; Dugo et al., 2004, 2006a,b,c; Mondello et al., 2005; Lı´sa et al., 2013). In addition to these two main types, some other solvent combinations have been also reported as well, such as toluene, hexane and ethylacetate (Schuyl et al., 1998), methanol and acetonitrile (van der Klift et al., 2008), and 13 different eluent systems containing hexane, heptane, or isooctane with the addition of acetonitrile, propionitrile or butyronitrile (Muller et al., 2006), heptane with acetonitrile or acetone (Macher and Holmqvist, 2001), or acetone with acetonitrile (Nikolova- Damyanova et al., 1995a,b).
Hexaneeacetonitrile mobile phases have a unique property of possible regioisomeric resolution of unsaturated TG, which has not been reported for chlorinated mobile phases. The disadvantage of a hexaneeacetonitrile system is the low solubility of acetonitrile in hexane, which is only about 1%e1.5% at ambient temperature (Adlof and List, 2004). The miscibility problem strongly limits the range of applicable chromatographic conditions and significantly contributes to the reproducibility problem, which can be partly solved by using continuous magnetic stirring. Two alternative approaches have been tested regarding how to solve the miscibility limitation, while maintaining excellent chromatographic resolution, including the resolution of positional isomers. When acetoni- trile is replaced by propionitrile, the miscibility is much better and the regioisomeric resolution is almost the same (Muller et al., 2006; Lı´sa et al., 2009a,b), but a serious health hazard arises because of the toxicity of propionitrile. The second approach relies on the addition of a third solvent with good mutual miscibility with both acetonitrile and hexane. The ideal combination is hexanee2- propanoleacetonitrile (Lı´sa et al., 2009a,b; Cvacka et al., 2006; Han et al., 1999; Holcapek et al., 2009, 2010; Vrkoslav et al., 2013), where remarkable improvements in the reproducibility of retention times are observed in comparison with traditional hexaneeacetonitrile binary mixtures (Muller et al.,
2006; Lı´sa et al., 2009a,b). Standard deviations of retention times for three selected peaks (PLP, PLL, and LLL) in hexanee2-propanoleacetonitrile are 0.4%, 1.0%, and 0.7% for one-day measurements compared to 7.4%, 6.8%, and 5.2% for a hexaneeacetonitrile mobile phase (Lı´sa et al., 2009a,b).
Some shifts in retention times can occur on a longer time scale, but they can be efficiently eliminated by the use of the relative retention factor, r¼(tR,TGtM)/(tR,stdtM). There are some important issues concerning mobile-phase preparation, which should be followed to obtain good reproducibility.
Mobile phases should be prepared fresh every day using solvents dried with molecular sieves and kept in tightly closed containers to avoid evaporation. A low percentage of additives in hexane should be premixed in solvent containers (Dugo et al., 2006a,b,c;Lı´sa et al., 2009a,b). The degassing of mobile phase by an automatic degasser is preferred over continuous stripping with a stream of helium or sonication. Columns are conditioned using a low flow rate of the initial gradient composition (50mL/min) overnight and the standard flow rate for 1 h before the analysis (Lı´sa et al., 2009a,b).
If the resolution is not sufficient on one silver-ion column, then more columns can be coupled in series, as demonstrated in several works (Adlof and List, 2004; Adlof, 1994, 1995; Lı´sa et al., 2009a,b;
Holcapek et al., 2009, 2010). An increased length of chromatographic column improves the resolution of critical pairs, for example regioisomeric doublets (Lı´sa et al., 2009a,b; Holcapek et al., 2009, 2010) or DB positional isomers of FA derivatives (Juaneda et al., 1994). Unlike in nonaqueous reversed- phase (NARP) systems, the back pressure is not a limiting factor here because mobile phases typically consist of low-viscosity organic solvents (e.g., hexane, dichloromethane, dichloroethane) with a low percentage of polar modifier. Limiting factors are mainly long retention times associated with the extended column length and peak broadening effects for the multiple column coupling.
3.3 TEMPERATURE
Temperature plays an important role in the optimization of the chromatographic separation of lipids, which is not limited to Ag-HPLC, but is valid for other separation modes as well, such as NARP (Holcapek et al., 1999, 2003, 2005; Lı´sa and Holcapek, 2008). Increased temperature can result in the loss of resolution for critical TG pairs in NARP mode, where the retention time depends inversely on the temperature. In the case of Ag-HPLC, the correlation among retention times, temperature, and chromatographic resolution is more complex. The basic rule for the temperature dependence in Ag-HPLC mode is that higher temperature means higher retention times (Adlof and List, 2004; Lı´sa et al., 2009a,b, 2013; Adlof, 2007), which is rather an unusual behavior not known for other HPLC modes, where just the opposite behavior is common. The magnitude of this effect is directly related to the DB number and is more evident withcis-DB (Adlof and List, 2004). A possible explanation for this strange behavior has been proposed by Adlof (Adlof and List, 2004) based on the different stability of the acetonitrile complex with silver ions, which is probably exothermic and thereby less stable at higher temperatures, which allows an increased number of interactions for the analyte with silver ions at higher separation temperatures, resulting in higher retention.
Within a certain range, a temperature decrease causes lower retention in hexaneeacetonitrile mobile phases, as demonstrated for FAME (Adlof, 2007) and TG (Adlof and List, 2004; Adlof, 2007).
The retention of FAME standards decreases with decreasing temperature from 20 to10C; however, it significantly rises with a further temperature decrease (20C). A similar trend is observed for TG, but the retention behavior changes at 0C. This behavior is not observed in chlorinated solvent systems (Adlof and List, 2004). The sudden increase in retention times at very low temperature can be
explained by several temperature-related factors (Adlof, 2007): (1) solubility of the sample in the mobile phase, (2) solubility of acetonitrile in hexane, (3) changes in the flexibility/3D configuration of the analyte or stationary phase. Another possible explanation is that the number of unsaturated mol- ecules coordinated in the complex with the silver ion depends on the temperature, whereas only one unsaturated molecule forms the complex at 25C in comparison to two coordinated molecules at 0C (Winstein and Lucas, 1938). Temperature gradients (Adlof, 2007) could be used for the optimization of chromatographic resolution of complex FAME or TG, instead of the more common solvent compo- sition gradient, but in our best knowledge this idea has not yet been used in any published paper.