Key words

1 Introduction

Only 3 years after the introduction of capillary electrophoresis (CE) by Jorgenson and Lukacs [1], the first paper on CE employing a nonaqueous electrolyte (tetraethyl ammonium perchlorate/hydrochloric acid in acetonitrile) was published [2]. From that time on nonaqueous capillary electrophoresis (NACE) was distinguished from aqueous CE by the use of background electrolytes (BGE ) based on purely organic solvents. This definition will also be followed in the present review. Some of the most convincing reasons for favoring NACE over CE with aqueous BGEs are [3, 4] as follows:

  • Improved solubility of large number of analytes .

  • Improved separation selectivity.

  • Lower electric current allowing the use of higher separation voltages.

  • Higher plate numbers.

Due to these features NACE has faced increasing interest over the last decade with more than 200 publications since 2004 (found in SciFinder). This is also reflected in several review papers discussing theoretical aspects [3–6] as well as listing applications of NACE [7–10]. Thereby, similar as in aqueous CE in most cases spectrophotometric detection is employed and only a fraction (less than one quarter) of NACE applications describe the use of mass spectrometric (MS) detection . There is only one review article so far, specifically dedicated to NACE–MS which was published by Scriba in 2007 [11]. Nevertheless, most review articles focusing on CE–MS in general also include sections dealing with NACE–MS [12–17].

1.1 BGE Systems for NACE

BGE s for NACE commonly consist of an electrolyte (either a salt and acid /base or mixtures thereof), additives, and a solvent or solvent mixture. In the subsequent sections, these BGE constituents will be discussed focusing on their role in NACE–MS with respect to both, separation and detection.

1.2 Solvents for NACE and NACE–MS

One main asset of NACE is the possibility to select from a large range of different solvents. Table 1 gives an overview of physicochemical properties of solvents frequently used in NACE in comparison with water. As can be seen from these data, relevant physicochemical properties vary substantially between solvents and it seems obvious that the choice of solvent can be a valuable tool for manipulating separations. Separation of analytes in electrophoresis is governed by differences in their electrophoretic mobility μ ep, i.e., their ability to migrate according to their ionic radius/charge ratio. These parameters (ionic radius and charge) are both influenced substantially by the type of solvent employed. Different solvents lead to changes in the size of the solvated ion thereby influencing its ionic radius ; dielectric constants ε and acid base properties of the solvent affect the degree of protonation/deprotonation and with it the charge of the analyte. In addition to that, also the viscosity (η) of the solvent determines migration velocities and subsequently the speed of separation. Actually the ratio of ε/η (given in Table 1) can be seen as good parameter for comparing solvents or solvent mixtures with respect to ion mobilities, whereby lower ε/η values imply slower migration of the ions [11]. An in-depth discussion of solvent effects in NACE would be beyond the scope of this book chapter, but more comprehensive information is available from several review articles [3, 4, 18, 19].

Table 1 Properties of solvents at 25 °C [9]
Full size table

Focusing on the situation in NACE–MS, when choosing an appropriate solvent not only factors related to separation have to be observed, but also the effect of the chosen solvent on the performance of the MS detector. When using a triaxial sheath flow interface (as done in the majority of NACE–MS applications published so far) the effluent from the separation capillary is substantially diluted by the sheath liquid (which will be discussed later), so the sprayed solution should mainly consist of the sheath liquid and conditions were thus supposed to be dominated by its composition. Nevertheless, the solvent used for NACE still influences the efficiency of the electrospray process and thereby important parameters like signal-to-noise (S/N) ratio and limit of detection (LOD). This has been discussed in more detail in two interesting papers, focusing on solvent properties and their role in detection in NACE–MS [19, 20]. Studies employing an organic/aqueous (isopropanol:water = 4:1) sheath liquid and BGE s based on several solvents (methanol, acetonitrile, dimethylsulfoxide, formamide, N-methylformamide, and N,N-dimethylformamide) revealed substantial differences in the LODs obtained. Only methanol and acetonitrile provided similar results for the tested analytes (2-aminobenzimidazole, procaine, propranolol, and quinine) as observed with aqueous electrolytes; for the other solvents less favorable LODs were recorded. In the case of 2-aminobenzimidazole, the LOD with formamide and N-methylformamide in the BGE was more than 300-fold higher than the one obtained with methanol, acetonitrile, or the aqueous BGE [19]. A consequence of these findings may be the fact that when browsing the applications listed in Table 2 the majority of NACE–MS applications is performed using BGEs containing methanol, acetonitrile, or mixtures of these two solvents .

Table 2 Applications of nonaqueous capillary electrophoresis–mass spectrometry (since 2004)
Full size table

1.3 Electrolyte Systems for NACE–MS

Focusing on electrolyte systems for NACE, the impact of the solvent employed has to be considered first. One major prerequisite is the solubility of the selected electrolyte system in nonaqueous solvent systems. Some of the most popular buffer systems in aqueous CE such as phosphate and borate are hardly soluble in nonaqueous media. Second, the critical parameter giving an idea about the dissociation of an electrolyte is the relative permittivity ε of the solvent. As can be seen from Table 1, for most solvents employed in NACE, ε is lower than in the case of water. As a consequence of this fact, comparing a given concentration of an electrolyte substance in water and in a nonaqueous solvent (such as methanol or acetonitrile) its conductivity will be lower in the nonaqueous medium. For this reason, relatively high concentration of electrolyte substances can be employed in NACE without reaching the limiting electrical current. This can clearly be seen from the applications listed in Table 2, where electrolyte systems containing 100 mM ammonium acetate and 40 % of acetic acid could be used without any problems with excessive current [21].

Comparing NACE with spectrophotometric detection and NACE–MS, it can be observed that using MS as detector has a clear impact on the selection of components that can be included in the BGE , whereby no significant difference exists between aqueous and nonaqueous electrolyte systems. The MS detector is far more prone to interferences caused by physicochemical properties of electrolyte ingredients than a photometric one. Unfortunately, the need to ensure compatibility with MS reduces the number of selectable electrolyte ingredients substantially. Focusing on MS with the most commonly used ionization technique, namely, electrospray ionization (ESI) [22] the use of volatile BGE components is mandatory. This prerequisite excludes a series of substances which are of very popular in CE, such as buffers based on phosphate or borate (although these two are not employed in NACE due to solubility issues in the commonly employed organic solvents), capillary coatings based on alkylammonium salts or sulfonate/sulfate additives. BGEs mostly used in CE–MS are based on formic acid , acetic acid (and their ammonium salts), carbonate, and solutions of ammonia or alkylamines if strongly alkaline conditions are needed. Here the situation in NACE–MS is quite similar. Nevertheless, there are some reports on the use of other BGE systems containing nonvolatile ingredients and their impact on ionization in ESI–MS exist [23]. In-depth information about requirements for MS-compatible CE electrolytes, in general, can be found in a comprehensive review article published by Pantuckova et al. [24].

1.4 pH in Nonaqueous Systems

A crucial parameter in optimizing the BGE for CE in general is the selection of the appropriate pH value, as this factor influences the direction and magnitude of the electro endoosmotic flow as well as the degree of protonation/deprotonation of the analytes and with it the direction and magnitude of their electrophoretic mobilities . Whereas measuring and adjusting the pH is a more or less simple task in aqueous systems, the concept of pH measurement and adjustment cannot be transferred easily to nonaqueous conditions. Several approaches to overcome this problem have been suggested so far [11]. One option is to measure pH in nonaqueous solvents employing the same procedure as in aqueous media. This leads to the so-called apparent pH a value that is quite often used to describe nonaqueous buffer systems. Nevertheless, it has to be taken into account, that this “apparent pH” only allows comparison between BGEs based on the same solvent. An alternative is pH calculation using the Henderson–Hasselbalch equation. This approach can only be used if the pK values of the employed acid or base in the respective solvent are available. A comprehensive discussion on the issue of pH measurement and adjustment in nonaqueous solvents has been published by Porras and Kenndler [6].

1.5 Capillary Coatings and Additives

Capillary coatings are quite popular in CE. They are used, for example, to reduce analyte/capillary wall interactions or to suppress or even reverse the electro endoosmotic flow, just to name a few reasons for their application [25]. Dynamic coating (i.e., the coating substance is added to the BGE ) and static coating (the coating is attached to the capillary wall either by covalent bonds or by strong electrostatic interaction) can be distinguished. When moving from aqueous conditions to NACE it has to be ensured that the coating is still stable even when purely organic BGEs are employed. Furthermore, most substances used for dynamic coating are not compatible with MS. So either capillaries with covalently bonded coatings are employed in NACE–MS [26] or substances like hexadimethrine bromide that show strong interactions with the capillary wall even under completely nonaqueous conditions [27].

The group of additives for CE comprises a wide range of quite different substances. These can be micelle-forming agents, ion-pairing reagents, chiral selectors , and so on just to name a few. Also in the case of additives analyte/additive interactions have to be reevaluated when moving from aqueous to nonaqueous media. Similar as in the case of substances employed for capillary coating, most of the additives commonly employed in CE are not compatible with MS detection . A strategy to combine NACE with nonvolatile additives such as cyclodextrins with MS detection is to select conditions where these additives migrate toward the capillary inlet [26, 28, 29]. So no disadvantageous suppression effects are encountered as the cyclodextrins are not reaching the MS ion source.

1.6 Sheath Liquids for Nonaqueous Capillary Electrophoresis–Mass Spectrometry

From the NACE–MS papers listed in Table 2, only in one case a sheathless interface is employed for CE–MS coupling [30]. In this study, a recently developed sheathless interface was compared to a conventional triaxial sheath flow interface with respect to its performance in NACE–MS of acid ic compounds. In all other studies, the addition of an appropriate sheath liquid is required for guaranteeing a stable electrospray. The sheath liquid serves as a makeup flow (to reach the minimum flow rates needed for the ESI source), it is needed to close the electric circuit of the CE system (as no second electrolyte vial is present in CE–MS coupling), and its composition should enhance ionization efficiency and probably overcome less favorable characteristics of the BGE (with respect to ionization). When comparing sheath liquid compositions used in aqueous CE–MS and those for NACE–MS not real difference can be detected. In most cases a mixture of water and an alcohol (mainly methanol or 2-propanol) or acetonitrile together with small amounts of a volatile salt (often ammonium acetate or formate) and/or a MS compatible acid or base (formic acid, acetic acid, or ammonia, respectively) are the best choice. An example demonstrating a further functionality of the sheath liquid was presented by Gomez-Caravaca et al. [31]. By adding 2.5 mM of NaOH to sheath liquid, sodium formate clusters were formed that could be directly used for mass calibration of the TOF instrument.

1.7 Optimization of NACE–MS Parameters

When searching for the best operational parameters for NACE–MS experiments, mutual interference between several experimental parameters like BGE composition, composition of the sheath liquid, sheath liquid flow rate, or nebulizer pressure has to be taken into account. Posch et al. have published an in-depth study discussing the possibility to use a design of experiments approach to study the influence of the background electrolyte on separation and detection in NACE–MS [20]. A scheme depicting potential interactions and influences is shown in Fig. 1. In this work it was proven that at high electro endoosmotic flow conditions, separation can be optimized without inferences from the MS detection system.

Fig. 1
figure 1

Visualization of the interactions and influences of the process parameters on the response variables. CE parameters are shown on top, MS parameters are shown below the response variables. Reproduced from [20] with permission

Full size image

2 Applications

Table 2 gives an overview of NACE–MS applications published since 2004. Applications are grouped according to the field of application and listed in chronological order. In this table, relevant information concerning the type of analytes investigated and/or the field of application, BGE and sheath liquid composition, and the type of MS instrument/ionization source used is provided. Additionally, remarks on characteristic features of the work are given.

2.1 NACE–MS of Plants and Natural Products

The analysis of plants and natural products is one of the main fields of application of NACE–MS. Unger et al. compared HPLC with CE and NACE for the analysis of crude extracts from Ancistrocladus species [21]. Although the highest number of resolved components was achieved by HPLC, NACE allowed the separation of cis/trans isomers (that could not be resolved using HPLC) whereas conventional CE with an aqueous electrolyte led to comigration of all analytes. Huhn et al. designed a CE system allowing simultaneous laser-induced fluorescence and MS detection [32]. Employing this setup, β-carbolines from an Ayahuasca sample were analyzed, whereby distinctly different migration orders were achieved in NACE compared to CE with an aqueous BGE . The group of Stuppner used NACE–MS with a BGE based on mixed solvents (acetonitrile/methanol = 9/1) for the investigation of several plant extracts [33–35]. A NACE–MS method for the analysis of alkaloids in tobacco and chewing gums was developed by Chiu et al. [36]. In their paper they investigated a series of nonaqueous BGEs differing in apparent pH, methanol/acetonitrile ratio, as well as type and concentration of electrolyte employed. The combination of NACE–MS and field amplified sample stacking for the high-sensitivity analysis of quinolizidine alkaloids was presented by Wang et al. [37]. Buchberger et al. achieved the separation of cinchona alkaloids extracted from cinchona bark [38]. Employing a rather complex nonaqueous BGE based on formic acid , acetic acid, and ammonium formate in a mixture of methanol/ethanol and acetonitrile, a series of diastereomeric compounds could be separated. The group of Huhn published a series of papers on the use of NACE–MS for the analysis of forensically interesting alkaloids from several plant species [20, 39, 40]. In one of these papers (as discussed earlier) the use of a “design of experiment” for optimization of NACE–MS parameters is discussed [20].

2.2 NACE–MS for Bioanalytical Applications

A second major field of application of NACE–MS is biomedical analysis. Anderson et al. described a NACE–MS method for the determination of lidocaine and its metabolites in human plasma [41]. Alamethicin peptides from Trichoderma viride were analyzed by Psurek et al. employing NACE–MS [42, 43]. Comparing the results obtained with aqueous and nonaqueous conditions revealed improved separation efficiency for the nonaqueous BGE as well as substantial selectivity changes. The latter might be attributed to changes in the shape of the peptide when switching from aqueous to nonaqueous conditions . Cyclodextrin -mediated NACE–MS was employed for the determination of salbutamol enantiomers in urine by Servais et al. [28]. The developed method allowed the baseline separation of the two enantiomers in less than 12 min. Due to their low solubility in aqueous media, phospholipids are not easily accessible to CE–MS analysis. Gao et al. investigated the potential of different nonaqueous BGE systems for the NACE–MS analysis of these compounds [44]. Morales-Cid et al. developed sophisticated automated instrumentation, including sample pretreatment steps such as packed sorbent microextraction and microdialysis for the analysis of drugs in body fluids by NACE–MS [45, 46]. Although both, aqueous and nonaqueous electrolytes were studied, the NACE approach was selected due to the increased sensitivity obtained. Employing the right mixture of methanol and acetonitrile allowed to adjust selectivity and to reduce analysis times. A series of antidepressants were separated by NACE and subsequently detected using MS by Sasajima et al. [47]. As can be seen from Fig. 2, great improvement in separation was achieved when moving from an aqueous BGE to a nonaqueous one. Interestingly, the BGE finally selected for this analytical problem can no longer be seen as a purely nonaqueous one, as it contains 1 % of water. Amphetamines in race-horse plasma were analyzed by Li et al. [48]. Fentanyl derivatives were separated using NACE by Rittgen et al. [49]. The analysis of these compounds gains more and more interest as clandestine fentanyl laboratories produce these substances for the illegal drug market. An interesting study comparing not only CE with aqueous and nonaqueous conditions, but also two different types of ESI interfaces for CE–MS coupling was published by Bonvin et al. [30]. As can be seen from Fig. 3, switching from an aqueous BGE to NACE substantially improved the resolution of the investigated test substances. In addition, sensitivity obtained with NACE–MS was 5–10 times better than in CE–MS, although also a substantially higher noise level was observed with the nonaqueous BGE.

Fig. 2
figure 2

Separation of 20 antidepressants—comparison between NACE and aqueous CE. BGE , (a) 50 mM ammonium acetate and 1 M acetic acid in acetonitrile, (b) 1 M formic acid in water; all other parameters are identical. Reproduced from [47] with permission

Full size image
Fig. 3
figure 3

CE–MS electropherograms of acidic compounds in negative ESI obtained for selected nonsteroidal anti-inflammatory drugs (dissolved at 1 μg/mL in ACN-MeOH 60:40 (v/v)) with the sheath liquid interface in (A) aqueous CZE mode; BGE : ammonium acetate 50 mM, pH 8.5 and (B) NACE mode; BGE: ammonium acetate 5 mM in ACN-MeOH 80:20 (v/v). (C) CE–MS electropherograms with the sheathless interface in NACE mode; BGE: ammonium acetate 5 mM in ACN-MeOH 80:20 (v/v). Peaks: Ind: Indomethacin; Sup Suprofen; Dic Diclofenac, Ibu Ibuprofen, Mef Mefenamic acid, Flu Flufenamic acid. Reproduced from [30] with permission

Full size image

2.3 NACE–MS in Food Analysis

Nonaqueous conditions are definitely favorable when it comes to the CE analysis of samples with low solubility in water. This fact has been exploited in two studies describing the NACE–MS analysis of olive oils and olive fruit with respect to phenolic compounds [31] and phospholipids [50]. β-Agonists in pork meat were analyzed by NACE–MS and HPLC–MS [51].

2.4 NACE–MS for the Analysis of Technical Products and Environmental Samples

Groom and Hawari investigated the formation of complexes including hexamethonium perchlorate (substances frequently used as rocket fuel) in both aqueous and polar nonaqueous solvents [52]. The resulting complexes were resolved employing NACE and detected by ESI–MS. A characteristic of this work is the rather unusual BGE based on a mixture of 2-propanol and acetone. Organic polymers are often insoluble in aqueous media. For this reason separation methods working in nonaqueous solution are preferable for the analysis of such samples. Simo et al. demonstrated the suitability of NACE–MS for the analysis of synthetic polymers (poly (N ε -trifluoroacetyl-l-lysine)) [53]. Thereby, structures containing up to 38 monomers could be resolved. Scrano et al. developed a NACE–MS method allowing the identification and quantitation of two novel degradation products originating from the photolytic reaction of oxasulfuron [54]. Another polymer-related application of NACE–MS has been published by Morin et al. [27]. The separation of the neutral polyethylene oxide surfactant was based on its complexation with ammonium in methanol as solvent. More than 25 oligomers of this surfactant could be characterized. Aqueous CE, open-tubular capillary electrochromatography and NACE, all coupled to MS, were compared with respect to their potential for the impurity profiling of drugs by Vassort et al. [55]. The results obtained within this study suggest that some of the previously developed CE–MS methods should be replaced by NACE–MS due to improved separation capabilities. Additionally, NACE appears attractive as a large portion of drug candidates are poorly soluble in water. Electrokinetic chromatography with cyclodextrins in nonaqueous media was employed for the analysis of various acidic drugs by Mol et al. [26, 56]. The effect of the cationic cyclodextrins on the ESI process was studied, whereby the separation voltage applied led to migration of these components in direction of the inlet vial, thereby not interfering with the ionization process. A fast NACE–MS method for the speciation of organotin compounds, substances commonly employed as antifouling agents, was presented by Malik et al. [57]. The use of a homemade CE instrument allowed applying separation voltages as high as 35 kV together with the use of short capillaries—thereby reducing analysis times to 2.5 min.

2.5 NACE–MS: Fundamental Investigations

Several papers discussing fundamental issues with respect to the coupling of NACE with MS have been published so far. Steiner and Hassel performed in-depth investigations on the influence of solvent properties on separation and detection [19]. They compared a series of solvents with respect to analysis times, separation efficiency, as well as performance in combination with ESI–MS detection . Some of the findings from this paper have already been discussed in the previous section on “Solvents for NACE and NACE–MS.” The potential of large volume sample stacking in combination with NACE–MS was investigated by Kim et al. whereby a 400-fold enrichment of anionic analytes was achieved [58]. Technical obstacles arising from the long sample matrix plug were solved by supplying a backup run buffer from the outlet vial of the CE system. Cyclodextrins are widely used in electrokinetic chromatography for selectivity manipulations. Unfortunately they can cause adverse effects in ESI–MS detection due to the occurrence of ionization suppression. Nonaqueous electrokinetic chromatography with either anionic [56] or cationic [26] cyclodextrins has successfully been coupled to MS detection . In a further paper, Servais et al. discussed the influence of BGE composition and type of cyclodextrin used on the detector response observed in cyclodextrin -mediated NACE–MS [28].

3 Materials

In this section instrumentation and materials for a typical NACE–MS application are discussed. Some of these points are also valid for normal CE–MS with aqueous BGE .

3.1 BGE for NACE–MS

BGE ingredients have to comply with both, requirements from NACE and requirements from ESI–MS. From the wide range of solvents employed in NACE with spectrophotometric detection only a few are also used in combination with MS: these are alcohols (methanol, ethanol) and acetonitrile. Electrolyte ingredients (salts, acids , or bases ) have to be soluble in the selected solvent and have to be compatible with ESI–MS. So in most cases low molecular weight organic acids (formic acid, acetic acid) and/or their ammonium salts are employed .

3.2 Sheath Liquid for NACE–MS

Sheath liquids used in NACE–MS are almost identical to those in aqueous CE–MS. Although purely nonaqueous sheath liquids can be used (methanol, propanol, or acetonitrile/alcohol mixtures) most sheath liquids in NACE contain 20–50 % water together with a small amount of acid /base or a volatile salt to enhance ionization.

3.3 CE Instrumentation

“7100 CE System“ (Agilent), Beckman PA 800 (SCIEX Separations) or equivalent, equipped with an ultraviolet (UV) absorbance detector, high voltage power supply up to +−30 kV, and autosampler for both hydrodynamic and electrokinetic injection. Also a special capillary cartridge for hyphenation with MS is needed. Due to the higher volatility of organic solvents (compared to aqueous BGE s) a cooling option for the tray, housing the sample and the electrolyte vials is advisable.

3.4 CE–MS Interface

The majority of CE–MS applications are performed using a triaxial sheath flow interface like the one available from Agilent (G1607A or G1607B). For supplying the sheath liquid ideally an HPLC pump with a 1:100 flow splitter is employed. A second option is the use of a syringe pump (e.g., from Harvard Apparatus, South Natick, MA, USA) whereby an increased baseline noise due to flow rate fluctuations must be taken into account.

3.5 MS Instrument

In most cases, MS instruments that offer commercially available dedicated interfaces for CE–MS coupling are preferable. Apart from that, MS instruments with an ionization source where the sprayer needle is grounded, whereas high voltage is applied to the MS orifice, as is the case in Agilent and Bruker instruments, for example, substantially facilitate CE–MS coupling. As CE is a highly efficient separation technique resulting in narrow peaks a sufficiently fast MS instrument is advantageous. In recent times, TOF and Q/TOF instruments have become the most frequently used instruments in CE–MS coupling.

3.6 Fused-Silica Capillaries

For example, from Polymicro Technologies (Phoenic, AZ) with inner diameter and outer diameter of 50 and 360 μm, respectively, and sufficient length to introduce the capillary into the MS interface . The capillary length can vary significantly due to the different layout of the available CE–MS systems and may be in the range between 60 cm and more than 100 cm. If coated capillaries are employed, stability of the coating in nonaqueous BGE s has to be ensured.