- Email: [email protected]
Preparation and characterisation of anion-exchange latex-coated silica monoliths for capillary electrochromatography
Journal of Chromatography A, 1109 (2006) 10–18
Preparation and characterisation of anion-exchange latex-coated silica monoliths for capillary electrochromatography Joseph P. Hutchinson a , Emily F. Hilder a , Miroslav Macka a , Nebojsa Avdalovic b , Paul R. Haddad a,∗ a
Australian Centre for Research on Separation Science (ACROSS), School of Chemistry, Faculty of Science, Engineering and Technology, University of Tasmania, Private Bag 75, Hobart, Tasmania 7001, Australia b Dionex Corporation, 1228 Titan Way, Sunnyvale, CA 94088-3603, USA Available online 8 September 2005
Abstract Silica monoliths coated with functionalised latex particles have been prepared for use in monolithic ion-exchange capillary electrochromatography (IE-CEC) for the separation of inorganic anions. The ion-exchange monoliths were prepared using 70 nm quaternary ammonium, anion-exchange latex particles, which were bound electrostatically to a monolithic silica skeleton synthesised in a fused silica capillary. The resulting stationary phases were characterised in terms of their chromatographic performance and capacity. The capacity of a 50 m diameter 25 cm latex-coated silica monolith was found to be 0.342 nanoequivalents and 80,000 theoretical plates per column were typically achieved for weakly retained anions, with lower efficiency being observed for analytes exhibiting strong ion-exchange interaction with the stationary phase. The electroosmotic flow (EOF) was reversed after the latex-coating was applied (−25.96 m2 V−1 s−1 , relative standard deviation (RSD) 2.8%) and resulted in anions being separated in the co-EOF mode. Ion-exchange interactions between the analytes and the stationary phase were manipulated by varying the ionexchange selectivity coefficient and the concentration of a competing ion (phosphate or perchlorate) present in the electrolyte. Large concentrations of competing ion (greater than 1 M phosphate or 200 mM perchlorate) were required to completely suppress ion-exchange interactions, which highlighted the significant retention effects that could be achieved using monolithic columns compared to open tubular columns, without the problems associated with particle-packed columns. The latex-coated silica monoliths were easily produced in bulk quantities and performed reproducibly in acidic electrolytes. The high permeability and beneficial phase ratio makes these columns ideal for micro-LC and preconcentration applications. © 2005 Elsevier B.V. All rights reserved. Keywords: Monoliths; Silica; Latex anion-exchangers; Capillary electrochromatography; Preconcentration
1. Introduction Monolithic stationary phases are an attractive choice for capillary electrochromatography (CEC). The in situ formation of a continuous stationary phase bed having a network of connected through-pores provides an efficient chromatographic medium with high permeability and favourable mass transfer characteristics. Furthermore, monolithic columns can be expected to show considerably higher ion-exchange capacities than open tubular columns due to their larger surface area. Monolithic columns also offer advantages over particle-packed columns. The monolith is formed in situ as a continuous bed within the capillary and
∗
Corresponding author. Tel.: +61 3 6226 2179; fax: +61 3 6226 2858. E-mail address: [email protected] (P.R. Haddad).
0021-9673/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2005.08.076
can be covalently bound to the capillary wall. This eliminates the problems associated with frits such as frit failure and bubble formation, which have often been encountered using particlepacked columns. Monolithic columns have also been shown to be much more permeable than packed capillaries due to the large through-pores formed [1,2] and this increase in permeability leads to shorter flush times, reduced back-pressure and frequently also to enhanced chromatographic performance of the column. The first uniform porous silica monolithic columns for HPLC were reported by Tanaka and co-workers [3] in 1996. In 2000, the sol–gel method was adapted to create silica monoliths in the capillary format and these were derivatised for reversedphase CEC applications [4,5]. In this sol–gel process, chloroand methoxysilanes undergo hydrolytic polymerisation in the presence of aqueous acid and poly(ethylene glycol) (PEG). The
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
silica network must bond to the silanol groups on the capillary wall to prevent shrinkage of the monolithic structure. If shrinkage occurs, voids between the monolithic structure and the capillary wall lead to reduced chromatographic efficiency or the monolithic material may be flushed from the column altogether. The macroporous morphology is formed when transitional structures during spinodal decomposition are fixed permanently by the sol–gel transition. Typically, the silica skeleton formed is 2 m thick and through-pores as large as 10 m can be obtained. Silica monolithic structures differ to those produced from organic polymers by virtue of their bimodal pore distribution and the presence of mesopores within the silica structure. These mesopores typically range from 1 to 50 nm and their size can be controlled by treatment with a base, such as ammonia, after formation of the macroporous silica skeleton. After fabrication, monolithic silica capillary columns have been typically derivatised using octadecyldimethyl-N,Ndiethylaminosilane to produce columns for reversed-phase applications [4,6]. Recently, Allen and El Rassi have investigated a variety of alternative derivatisation procedures for the separation of analytes using capillary electrochromatography. These included hydrophillic silica monolithic stationary phases bearing surface-bound cyano groups for the separation of polar species [7] and amphiphillic silica monoliths possessing surfacebound cationic octadecyl moieties [8] for the separation of neutral and charged species. Other advances in silica monolithic columns prepared for CEC include those with embedded particles [9,10], photochemical polymerisation [11], cationic sol–gel precursors to reverse the EOF [12], derivatisation with chiral selectors [13], sol–gel encapsulation of chiral proteins [14] and monoliths coated with ion-exchange materials [15,16]. Silicabased monoliths for CEC have recently been reviewed in detail by Allen and El Rassi [17]. To date, anion-exchange CEC using monolithic silica columns has received very little attention. The only documented examples in the literature are by Breadmore et al. In the first example [15], monolithic silica capillary columns were fabricated and coated with the soluble cationic polymer poly(diallyldimethylammonium chloride) (PDDAC). The resulting columns were used for the separation of inorganic anions. More recent work by the same authors was performed using dextran sulfate in between multiple PDDAC coatings to form polyelectrolyte multilayers on the monolithic silica surface. The subsequent columns were used for the separation of peptides and inorganic anions [18]. In the present study, monolithic silica columns were prepared in 50 and 75 m I.D. fused-silica capillaries and then coated with latex nanoparticles functionalised with quaternary ammonium groups (AS5A, Dionex). The positively charged latex particles adhere electrostatically to the negatively charged silanol groups on the silica monolith and create an anionexchange stationary phase suitable for ion-exchange capillary electrochromatography (IE-CEC). This approach has been used previously in our group where latexes have been coated onto the wall of fused-silica capillaries and used for open-tubular CEC and preconcentration techniques [19–22]. The stability
11
of the AS5A-coated monolith was tested and the reproducibility and chromatographic performance of the coated monoliths were assessed. This work is part of an on-going investigation to fabricate novel ion-exchange stationary phases to perform separations in CEC and also micro-ion chromatography formats. The ion-exchange selectivity of high capacity columns can be manipulated for faster and more efficient separations and will allow the determination of analytes in real samples, which cannot be analysed without preconcentration methods being performed.
2. Experimental 2.1. Instrumentation and physical characterisation of monoliths The capillary electrophoresis instrument used was an Agilent 3D CE (Waldbronn, Germany). Separations were carried out using 50–100 m I.D. fused-silica capillaries (Polymicro Technologies, Phoenix, AZ, USA) containing a monolithic silica bed with a length of 25 cm, unless otherwise stated. A 75 m I.D. 9.6 cm open-tubular AS5A-coated capillary was attached to the capillary containing the monolith using a 1 cm PTFE sleeve as described previously [23]. A detection segment was formed in this section of capillary by burning a window 1.1 cm away from the end of the monolith and 8.5 from the anode. This allowed UV detection to occur without problems of light scattering on the monolith, which would otherwise be a problem if detection through the monolithic structure was attempted. Anions were separated in the co-electroosmotic flow (co-EOF) mode where both the analyte anions and the EOF migrated towards the anode. The EOF was measured using acetone as a marker. The voltage used for the separations was varied between −1 kV and −30 kV, depending on the concentration of the background electrolyte (BGE) and the temperature was maintained at 25 ◦ C. Injection of anions was performed by hydrodynamic injection at the cathodic end. Scanning electron microscopy (SEM) images were obtained using an ElectroScan ESEM2020 Environmental Scanning Electron Microscope (Wilmington, MA, USA) operated in high vacuum mode using an accelerating voltage of 15 kV. Secondary electrons were detected using an Everhard–Thornley scintillator-type electron detector. The monolithic capillary was cut into 1 cm fragments that were mounted perpendicularly to a 12 mm pin-type aluminium stub using epoxy resin. Highresolution images were obtained by coating the capillary with gold (nominally 40 nm thick). In parallel with synthesis of the monolith in the capillary, synthesis was also carried out in a glass vial of larger volume in order to obtain a sufficient amount of the monolith for the determination of the porous properties. The pore volume and pore size distribution were determined using an Autopore III 9400 mercury intrusion porosimeter (Micromeritics, Norcross, GA, USA). The surface area was calculated from the BET isotherms of nitrogen adsorption and desorption using an ASAP 2010 (Micromeritics).
12
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
2.2. Chemicals AS5A latex (quaternary ammonium functionalised) particles were supplied as an 11% w/v suspension by Dionex (Sunnyvale, CA, USA). Standard solutions of 10 mM Br− , I− (Sigma–Aldrich, Milwaukee, WI, USA), Cl− (May and Baker, Footscray, Australia), BrO3 − , IO3 − , SCN− , CrO4 2− (Ajax, Melbourne, Australia) NO3 − and S2 O3 2− (BDH, Kilsyth, Australia) were prepared from sodium or potassium salts of analytical reagent grade. All samples were prepared in water obtained from a Milli-Q (Millipore, Bedford, MA USA) water purification system. Stock solutions of 100 mM 2-(Nmorpholino)ethanesulfonic acid (MES) and histidine (His) were prepared by dissolution in Milli-Q water of analytical grade reagents purchased from Aldrich. These were combined to form the 50 mM MES/His buffer. Phosphate at pH 7.0 and borate at pH 9.3 were diluted from 50 mM solutions purchased commercially (Agilent, Waldbronn, Germany). All BGEs used were filtered through 0.2 m filters (Millipore) and degassed under vacuum prior to use. 2.3. Procedure for creating latex monoliths The monolithic silica columns were prepared using the method outlined by Ishizuka [24]. Two milliliters of 99% purity tetramethoxysilane was added to 0.44 g PEG (MW = 10 000) and 0.45 g urea in 5 mL 0.01 M acetic acid. All sol–gel reagents were purchased from Sigma–Aldrich (Milwaulkee, WI, USA). The reagents were stirred at 0 ◦ C until a homogenous mixture formed (40 min) and this was then injected into the capillary using highpressure on the Agilent CE instrument and maintained at 40 ◦ C overnight. The capillary was then placed in an oven for 3 h at 120 ◦ C to complete the formation of the mesopores. The monolith was flushed with water and methanol and placed back in the oven at 330 ◦ C for 25 h to remove any residual organic material within the capillary. Finally, the monolithic silica column was flushed with water and methanol prior to latex coating. The monolithic silica columns were then treated with a tenfold aqueous dilution of the AS5A latex suspension provided by Dionex, which had been filtered through a 0.45 m Gelman nylon filter. Ten column volumes of the diluted AS5A latex suspension were flushed through the monolith and the effluent monitored using the UV detector on the instrument. Breakthrough of the latex was evident as a sharp increase in absorbance and was used to indicate that all available sites on the monolith had been saturated with latex. The column was then flushed with water and finally with at least 10 column volumes of BGE prior to use. Between runs the column was flushed with BGE prior to injection. 3. Results and discussion 3.1. Characterisation of latex-coated silica monolithic capillary columns To assess whether the AS5A latex particles had adhered to the silica surface of the monolith, the EOF prior to and after
Table 1 Comparison of the EOF in AS5A-coated open-tubular capillary columns, AS5Acoated monolithic silica capillary columns, and uncoated monolithic silica capillary columns EOF mobility (×10−9 m2 V−1 s−1 ) 50 m I.D. AS5A-coated monolith 1 50 m I.D. AS5A-coated monolith 2 75 m I.D. AS5A-coated monolith 50 m I.D. uncoated monolith 75 m I.D. uncoated monolith 75 m I.D. AS5A-coated OT capillary
−21 ± 2 (n = 11) −26.0 ± 0.1 (n = 13) −33.9 ± 0.3 (n = 3) 29.0 (n = 1) 33.4 (n = 1) −28.7 ± 0.6 (n = 3)
Conditions: columns: all columns were 33.5 cm in length with detection 8.5 cm from the end. BGE: 10 mM perchloric acid; voltage: −30 kV; temperature: 25 ◦ C; EOF marker: acetone.
coating with latex was measured. Table 1 shows that the EOF was reversed after coating to a value similar in magnitude to that reported for an open-tubular capillary coated with AS5A latex particles [15]. The EOF is a result of the bulk flow of ions in the diffuse layer formed at the stationary phase surface to maintain charge balance. This layer is called the electrical double layer and has an associated potential difference known as the zeta potential. Breadmore et al. [15] have previously investigated the dependence of the EOF in a silica monolithic column on the ionic strength of the BGE. They found that the EOF in silica monoliths increased with ionic strength, contrary to the behaviour of an empty capillary. This was explained by Breadmore et al. [15] to be due to the limited EOF in the small mesopores present in the silica skeleton at low ionic strength. Under such conditions, the thickness of the electrical double layer was sufficiently great that overlap from opposing sides of the pore occurred. However, when the ionic strength of the BGE was increased, the thickness of the electrical double layer formed at all surfaces was decreased, reducing the overlap of the electrical double layer within the mesopores. This allowed the mesoporous surface area to contribute to the overall EOF of the column. Typically in an empty capillary, increasing the ionic strength of the BGE decreases the zeta potential, which in turn reduces the magnitude of the EOF. The EOF in the latex-coated monoliths created in this work behaved similarly to the open-tubular columns in that it decreased with increasing ionic strength. This may be a result of the mesopores being obscured by the larger latex particles, thereby reducing the mesoporous contribution to the EOF thought to cause the deviation from empty capillary theory in terms of the velocity of the EOF. Upon coating the silica monolith with the AS5A latex particles, the time taken to flush the column with BGE was reproducibly 30% longer than for the uncoated monolith. The adsorbed latex particles clearly reduced the permeability of the column by decreasing the through-pore diameter. In the coating procedure, approximately six column volumes of a 1% latex suspension were needed to fully coat the monolith, indicating a considerable amount of the latex was retained on the column. The reversal of the EOF in the coated column was further evidence that there was very little bare silica remaining after the coating process.
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
13
Fig. 1. SEM pictures of the: (a) uncoated silica monolithic structure and (b) the AS5A-coated silica monolithic structure.
Several monoliths were viewed using scanning electron microscopy. Fig. 1 shows the structure of the 50 m I.D. AS5Acoated silica monolith in comparison to an uncoated silica monolith. The appearance of the latex-coated monolithic surface is not markedly different to that of the uncoated monolith and an SEM capable of greater resolution would be required for visual identification of the latex particles. However, the size and structure of the through-pores and the relatively thin silica skeleton could be seen clearly. The bimodal silica skeleton (approximately 2 m in width as measured from Fig. 1), is susceptible to decomposition at alkaline pH. This is undesirable as the latex coating would be removed and fragments of the silica skeleton may be dislodged, reducing the permeability and altering the integrity of the electrical circuit established during electrophoresis. The effect of alkaline BGEs was investigated by flushing the column with electrolytes at different pH and evaluating the EOF in successive runs. It was found that an EOF of significantly smaller magnitude occurred after approximately 7 runs at alkaline pH using tris(hydroxymethyl)aminomethane (Tris), phosphate or borate buffered electrolytes. However, when acidic electrolytes such as perchloric, acetic or phosphoric acid between pH 1–3 were used, the EOF was not altered significantly and a single column could be used for a period of several months. For this reason, only acidic BGEs were used in the remainder of this work. Measurement of the EOF was deemed to be the best method for determining the reproducibility of column fabrication since small differences in surface charge markedly alter the magnitude of the EOF. The average EOF and relative standard deviation (RSD) over 15 consecutive runs was found to be −25.96 × 10−9 m2 V−1 s−1 and 2.8%, respectively. A 2 m length of monolithic silica was synthesised in 50 m I.D. fusedsilica capillary and consecutive portions were cut and coated with AS5A latex particles in an identical fashion. These columns were used for testing the repeatability of the coating procedure and for comparing the performance between the columns.
The RSD between columns made from the same batch was 8% (conditions: voltage: −10 kV, BGE: 10 mM HClO4 , temperature 25 ◦ C, EOF marker: acetone) and was similar to that found between separate batches (7.4% RSD). Mercury intrusion porosimetry and BET adsorption isotherms were used to assess the physical properties of the silica monoliths used in this work. These techniques were performed on both uncoated and latex-coated bulk polymerisation samples, which were formed under the same thermal conditions as the monolithic capillary columns. The silica monoliths created were of a bimodal pore structure possessing macropores of 2.2 m in diameter and also mesopores within the silica skeleton of 8.3 nm diameter. The resulting material had a porosity of 70.4%. The surface area was increased from 167 to 186 m2 /g upon coating the silica monolith with latex nanoparticles as measured by BET adsorption isotherms. The behaviour of the monolithic columns under hydrodynamic flow conditions was assessed prior to their use in CEC. Hydrodynamic flow is used for the preconditioning of the column and for quantifying injection volumes into the column. The time required for elution of a neutral, hydrophilic (unretained) marker from the 50 m I.D. AS5A-coated monolith was measured at applied pressures ranging between 950 mbar and 12 bar. Elution was very rapid (<50 s) at pressures exceeding 6 bar, indicating that the columns were well suited to the pressure limitations of the Agilent CE instrument used in this work. 3.2. Ion-exchange capacities of the latex-coated monolithic columns The ion-exchange capacities of the AS5A-coated monolithic silica columns were measured by an adsorption/elution method. Solutions of different anions (10 M) were loaded onto the column hydrodynamically until saturation was observed. Water was flushed through the column to remove any residual anions not bound to the latex. Anions bound to the stationary phase were
14
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
Fig. 2. Ion-exchange capacities of two 50 m I.D. AS5A-coated monolith duplicates for a range of anions and a range of AS5A-coated open-tubular columns of varying diameter.
then eluted using 100 mM perchlorate and the elution monitored with the UV detector on the instrument. The normalised peak area was compared to calibration curves constructed for each anion. Fig. 2 compares the capacity of a latex-coated monolithic column with that of a latex-coated open-tubular column where the capacities shown are the average of the values obtained using six different inorganic anions. There was a large variation in capacity between the two monolithic columns evaluated, despite the fact that they were cut from consecutive portions of the same larger monolithic column and were coated in an identical manner. This suggested that the monolithic structure was not homogenous, although both columns had similar flow-rates and it could be seen under an optical microscope that the monolithic structures formed at the ends of each column were similar. This highlights the difficulty in trying to synthesise reproducible columns, even from the same batch. Theoretical calculations of ion-exchange capacity were performed for the latex-coated monolithic columns and the latexcoated open-tubular columns. Data for the capacity and size of the monodisperse latex particles were provided by Dionex (12 equiv./g and 70 nm, respectively) and the volume of latex coated within the column was calculated by assuming a monolayer of nanoparticles adsorbed to the available surface area. The formation of a monolayer of latex particles has previously been shown by scanning electron microscopy in the case of OTcoated columns [19]. To simplify the calculation, each latex particle was assumed to have the shape of a cube. The theoretical capacity of the monolithic columns was calculated on the basis of a monolayer coverage of latex particles over the monolithic surface area experimentally measured using BET adsorption isotherms. Values obtained for the OT columns are included in Fig. 2 and in all cases were of the same order, but somewhat higher, than the measured value. In the case of the latex-coated monolithic silica column the theoretical capacity was calculated to be approximately 250 times larger than an open tubular latex-coated capillary of the same dimensions. In a related study, Liu et al. [25] calculated that an avidin-coated 50 m I.D. silica monolith had a capacity that was 120 times greater than an open-tubular column of the same dimensions. However, the experimental value of the capacity of the latexcoated monolith was approximately 13 times higher than its
Fig. 3. Breakthrough volumes for 1 mM solutions of a range of analytes on a 50 m I.D. 25 cm AS5A-coated monolith.
open-tubular counterpart in one column and 5.5 times for the second column. This suggested strongly that a monolayer of latex particles had not formed at the monolithic surface. This may be attributed to a lower density of surface silanol groups on the monolith or it may be related to the physical morphology of the material, which might impede monolayer adsorption of the spherical latex particles. Breakthrough studies were performed by pumping a solution of the desired analyte through the capillary at various flow-rates and monitoring the UV absorbance of the effluent. Fig. 3 shows the results obtained and confirms that the breakthrough volume increased with increasing ion-exchange selectivity coefficient for the analytes tested. Moreover, it is evident that the magnitude of the flow-rate used had little effect on the breakthrough volumes obtained due to the improved mass transfer characteristics of the monolithic column. The random nature of the through-pores in the monolith increases the tortuosity of flow and improves mass transfer by reducing the distance of diffusion to the stationary phase, thereby aiding effective adsorption of the analytes. This indicates that these columns may be suitable for on-line preconcentration purposes. 3.3. Separation efficiency of the latex-coated monoliths Variation in the separation voltage resulted in changes in the linear velocity of the EOF. Van Deemter plots for a variety of analyte anions in a perchlorate BGE on two AS5A-coated monoliths were constructed and Fig. 4 shows these plots for bromide at different linear velocities. The minimum value of plate height (approximately 25 m) was observed at a linear velocity of about 1 mm/s, which corresponded to an applied voltage of approximately −15 kV. This represented an efficiency of 40,000 theoretical plates/m. This value was influenced by the masstransfer effects due to the existence of ion-exchange interactions between the analyte and the stationary phases, and also the fact that the system used involved the coupling of two capillary segments (i.e. the separation segment and the detection segment). The latter effect has been shown to considerably reduce the efficiency [23]. Peak efficiency was also measured as a function of the ionic strength of the BGE at constant voltage and the results are given in Table 2. A low voltage was required to avoid the effects of Joule heating as the BGE concentration was increased to 200 mM to reduce the ion-exchange interaction of iodide,
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
15
Table 2 Effect of BGE concentration on separation efficiency [ClO4 − ] in BGE (mM)
Theoretical plates per metre Br−
1 5 10 20 40 80 160 200
20,980 20,240 50,650 72,820 77,680 74,900 77,610 36,740
NO3 −
I−
30,820 19,600 50,430 61,020 34,420
1,320 1,820 1,200 3,040
35,980 8,790
27,420 7,840
a
a
SCN−
S2 O3 −
4,720 4,380 6,640 34,670 31,730 13,310 10,740
30,260 61,820 26,520
Conditions: column: 25 cm 75 m I.D. AS5A-coated silica monolith attached to a 10 cm 75 m I.D. open-tubular AS5A-coated column with a window at 8.5 cm via a PTFE sleeve; injection: 40 mbar for 40 s. Voltage: −1 kV; temperature: 25 ◦ C. a NO − and I− were co-eluted with this electrolyte. 3
thiocyanate and thiosulfate. In Table 2 it can be seen that as the ion-exchange interaction is decreased (by increasing the ionic strength of the BGE), the peak efficiency is increased. For example, bromide had an efficiency of approximately 21,000 plates/m when separation was performed using 1 mM perchlorate in the BGE and the efficiency increased to 80,000 plates/m when using 160 mM perchlorate in the BGE. The reason the efficiency was seen to be reduced at 200 mM was due to the reduction of the zeta potential and thus the magnitude of the EOF at high ionic strength.
mide, as indicated by the fact that its effective mobility was still increasing in the negative direction. On the other hand, when perchlorate was used as the competing ion (Fig. 5b) only 40 mM was required to limit the ion-exchange interaction of bromide with the stationary phase during the separation, while higher concentrations were needed to achieve the same goal for analyte ions of higher ion-exchange selectivity, such as thiocyanate. The separation selectivity could be controlled by the nature and concentration of the BGE, but it should be noted that increasing the ionic strength of the BGE increased the current during electrophoresis and necessitated the use of a lower applied voltage
3.4. Variation of electrolyte composition The chromatographic performance of the latex-coated silica monoliths was assessed by altering the composition and concentration of the BGE, predominantly with a view to manipulating the extent of the ion-exchange interactions between the analytes and the stationary phase. Fig. 5 shows the retention effects occurring in phosphate and perchlorate BGEs. In Fig. 5a, the retention of analyte anions having low ion-exchange selectivity coefficients are shown for phosphate BGEs of varying concentration. It can be seen that even at a concentration of phosphate of 1000 mM, ion-exchange effects were still apparent for bro-
Fig. 4. Van Deemter plots prepared using bromide as analyte at a range of applied voltages for two latex-coated silica monolithic columns. Conditions: columns: 25 cm 50 m I.D. AS5A-coated monolith attached via PTFE sleeve to 9.6 cm 75 m I.D. AS5A-coated open-tubular detection segment with window 8.5 cm from end; BGE: 5 mM perchloric acid; Injection: 50 mbar 5 s of 1.4 mM Br− ; temperature: 25 ◦ C.
Fig. 5. Selectivity effects for a AS5A-coated silica monolith using: (a) phosphoric acid BGE and (b) perchloric acid BGE Conditions: column: 50 m I.D. 24.1 cm AS5A-coated silica monolith attached via PTFE sleeve to 9.6 cm 75 m I.D. AS5A-coated open-tubular column with detection window 8.5 cm from end; Injection: 40 mbar 40 s of 1 mM solutions of indicated anions. Voltage: between −10 and −1 kV as required to reduce Joule heating; temperature: 25 ◦ C.
16
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
3.5. Latex-coated monolithic columns for in-line sample preconcentration
Fig. 6. IE-CEC separations of inorganic anions on a latex-coated silica monolith when the concentration of perchlorate as competing anion was varied. Detection was performed on-column through the monolithic bed. Conditions: column: 33.4 cm × 50 m I.D. AS5A-coated silica monolith with detection on-column 8.5 cm from anodic end; injection: 50 mbar 20 s of 2 mM solutions of the indicated anions; BGE: Tris/ClO4 − at varying concentrations; voltage: −15 kV; temperature: 25 ◦ C; UV detection: 195 nm.
to avoid Joule heating effects. This in turn lowered the speed of the separations achieved. The replicate 50 m I.D. AS5A-coated monolith (column 2) was also evaluated for its chromatographic performance and it exhibited the same selectivity effects as column 1, but with slightly lower retention as expected from its lower ion-exchange capacity. It was anticipated that the increase in the phase ratio when using latex-coated monolithic columns compared to latex-coated open-tubular columns would improve separation efficiency. In open-tubular CEC separations it is known that the capillary internal diameter should be less than 25 m to provide optimised efficiency as Taylor dispersion (resistance to mass transport in the mobile phase) increases steeply with column diameter. The monolithic stationary phases used in this work gave better efficiencies than those found in previous studies using AS5A latex-coated open-tubular columns [19,26], which achieved efficiencies of about 10,000 plates per metre. However, as stated earlier there are several reasons to believe that the efficiencies achieved in this present work may be less than optimal due to the coupling of capillaries. For example, the efficiency was found to be much greater when detection was performed through a 1 cm gap in the monolithic bed rather than by coupling a detection capillary segment. This can be formed by injecting a small amount of air into the capillary between the sol–gel solution prior to heat treatment and formation of the monolithic structure within the capillary. Such separations performed in this way using perchlorate BGEs of varying ionic strength are shown in Fig. 6.
In addition to the CEC separations described above, a further potential use of the latex-coated monoliths is for in-line sample preconcentration. Here, a short capillary segment containing the monolith (the preconcentration segment) is coupled to a longer segment of open capillary (the separation segment), in which the separation and detection are performed. This approach is similar to that used in a previous study in which we utilised latex-coated polymeric monoliths for in-line sample preconcentration [27]. It was necessary for the separation capillary to be coated with a positively charged polymer (PDDAC) in order to reverse the EOF so it was in the same direction as that for the preconcentration column. This provided a fast separation of anions in the co-EOF mode and minimised disturbances of the compressed analyte zone as it left the preconcentration segment and entered the separation segment. Fused-silica capillary of 50 m I.D. was found to be the best choice for creating the latex-coated silica monolithic preconcentration columns. Use of such relatively small internal diameter capillaries minimised shrinkage of the monolith from the inner wall of the capillary, which was a problem in larger diameter capillaries. Using smaller diameter preconcentration capillaries also resulted in lower operating currents and reduced the effects of Joule heating. Decoupling the preconcentration segment and the separation segment allowed separate optimisation of the dimensions of each segment. Detection limits can be improved since a larger diameter separation capillary can be used, which improves detection by providing a longer optical path-length. After sample loading, the bound analyte ions were eluted from the preconcentration segment and focused into a compact band using a transient isotachophoretic gradient, as described previously [21]. A fluoride/5-sulfoisophthalate discontinuous electrolyte system was used for this purpose. Upon the application of voltage the preconcentrated anions migrate in front of the 5-sulfoisophthalate strong electrolyte (SE) entering the preconcentration segment and when these ions reach the separation segment, ion-exchange interaction ceases and an electrophoretic separation occurs in the fluoride weak electrolyte (WE) filling the separation capillary. Fluoride has an electrophoretic mobility close to those of the preconcentrated analyte anions and this minimises the effects of electromigrational dispersion and produces satisfactory peak shapes for these analytes. 5-Sulfoisophthalate is an ideal SE as its ion-exchange selectivity coefficient is 14,000 times larger than that of fluoride and it has a suitably low electrophoretic mobility of −33.77 × 10−9 m2 V−1 s−1 , allowing a large range of anions to migrate electrophoretically in front of the isotachophoretic gradient formed. Fig. 7 shows the separation of anions obtained using this method, where the concentration of the fluoride WE was adjusted from 1 mM (which allowed binding of a range of ions on the preconcentration segment) to 100 mM (which permitted only strongly retained anions like chromate and iodide to be preconcentrated). It should be pointed out that part of the preconcentration procedure [21] involved forward-flushing of the monolith with WE after sample loading. Depending on the concentration
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
17
4. Conclusions
Fig. 7. Ion-exchange preconcentration and elution using a transient isotachophoretic gradient where the WE was introduced past the preconcentration zone using: (a) 1 mM F− and (b) 100 mM F− . Conditions: column: 25 cm 50 m I.D. AS5A-coated monolithic silica column coupled via PTFE sleeve to 10 cm 75 m I.D. PDDAC-coated open-tubular column with UV detection at 8.5 cm from end; WE: NaF used at concentrations of (a) 1 mM and (b) 100 mM and introduced from cathodic end for 1.1 column volumes; SE: 5-sulfoisophthalate was used at matching concentration to the WE used; injection: 6 bar 4 min of 1 M Br− , NO3 − , I− , SCN− , IO3 − and BrO3 − in water; voltage: −10 kV; temperature: 25 ◦ C; UV detection: 195 nm.
of the WE, this step can cause some analytes to be flushed to waste. This effect is apparent in Fig. 7b, in which the 100 mM fluoride WE has eluted the weakly retained analytes to waste and they therefore do not appear in the final chromatogram. The recoveries for chromate and iodide in Fig. 7b were 97 and 91%, respectively. A preconcentration factor of approximately 480 was achieved in this study for iodide when 4.8 capillary volumes of the sample were introduced into the preconcentration column compared to a standard CE injection occupying 1% of the column. In comparison, when latex-coated monolithic polymer stationary phases were used, a preconcentration factor of approximately 200 was achieved for the method used to preconcentrate iodide in seawater. However, in the previous study [27], the detection limit was much lower for iodide due to the analyte focusing achieved when using a latex-coated OT separation column coupled to the preconcentration column. The capacity of both the latex-coated polymer (17.1 picoequivalents/cm for a 75 m I.D. column) [27] used previously and the silica monoliths (13.8 picoequivalents/cm for a 50 m I.D. column) used in this study were found to be similar. However, it was possible to synthesise polymer monolithic phases in larger I.D. capillaries (250 m), which increased the volume of stationary phase of the preconcentration column and hence the effective capacity to a value of 386 picoequivalents/cm. The performance of this system could be enhanced through the use of a higher capacity stationary phase, which would be achievable by manipulating the pore size of the monolith or using latex particles of higher ion-exchange capacity.
There is a significant increase in the retention of analytes in CEC using latex-coated monolithic columns compared to their latex-coated open-tubular counterparts. This increase in retention of analytes can be attributed to the higher phase ratio and higher ion-exchange capacity of monolithic columns. However, the monolith stationary phases also required the use high ionic strength electrolytes in order to vary selectivity by manipulating the extent of ion-exchange interactions between the analytes and the stationary phase. This in turn led to problems with Joule heating effects and lower separation efficiencies for highly retained analytes. There were two disadvantages associated with using latexcoated monolithic silica columns. First, the silica backbone imposed a requirement for low pH electrolytes and its fragile nature could lead to blockages and reduced lifetime of the column. Second, there was a restriction in the choice of the capillary internal diameter used to house the monolithic columns because of shrinkage of the monolithic structure away from the inner wall occurring in larger internal diameter capillaries. Preconcentration of a range of inorganic anions was performed on a latex-coated monolithic silica preconcentration segment. Preconcentrated anions were eluted using a transient isotachophoretic gradient and separated electrophoretically when coupled with a separate segment of capillary designed for this purpose. However, the ion-exchange capacity of the preconcentration bed needs to be increased to enable reproducible preconcentration of anions possessing a wide range of IE selectivity coefficients and for the retention of strongly retained anions in high ionic strength matrixes. Latex-coated silica monolithic stationary phases are highly permeable compared to particle-packed columns and are of higher capacity than OT columns. These characteristics make such columns suitable for analyte preconcentration in real samples, which challenge the detection limit of traditional detection systems. Possible applications which will be focused upon in future work using latex-coated monolithic columns include the determination of trace contaminants such as (perchlorate) in drinking water and analysing trace inorganic anions in difficult matrices such as the determination of iodide and thiocyanate in seawater samples. Acknowledgements Financial support from the Australian Research Council is gratefully acknowledged. The authors would also like to thank David Steele of the Central Science Laboratory at the University of Tasmania for the SEM images. References [1] N. Vervoort, P. Gzil, G.V. Baron, G. Desmet, Anal. Chem. 75 (2003) 843. [2] N. Tanaka, H. Kobayashi, N. Ishizuka, H. Minakuchi, K. Nakanishi, K. Hosoya, T. Ikegami, J. Chromatogr. A 965 (2002) 35. [3] H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka, N. Tanaka, Anal. Chem. 68 (1996) 3498.
18
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
[4] N. Ishizuka, H. Minakuchi, K. Nakanishi, N. Soga, H. Nagayama, K. Hosoya, N. Tanaka, Anal. Chem. 72 (2000) 1275. [5] N. Ishizuka, K. Nakanishi, K. Hirao, J. Sol–Gel Sci. Technol. 19 (2000) 371. [6] N. Tanaka, H. Nagayama, H. Kobayashi, T. Ikegami, K. Hosoya, N. Ishizuka, H. Minakuchi, K. Nakanishi, K. Cabrera, D. Lubda, J. High Resolut. Chromatogr. 23 (2000) 111. [7] D. Allen, Z. El Rassi, J. Chromatogr. A 1029 (2004) 239. [8] D. Allen, Z. El Rassi, Analyst 128 (2003) 1249. [9] M.T. Dulay, R.P. Kulkarni, R.N. Zare, Anal. Chem. 70 (1998) 5103. [10] C.K. Ratnayake, C.S. Oh, M.P. Henry, J. High Resolut. Chromatogr. 23 (2000) 81. [11] M.T. Dulay, J.P. Quirino, B.D. Bennett, M. Kato, R.N. Zare, Anal. Chem. 73 (2001) 3921. [12] J.D. Hayes, A. Malik, Anal. Chem. 72 (2000) 4090. [13] J. Kang, D. Wistuba, V. Schurig, Electrophoresis 23 (2002) 1116. [14] M. Kato, K. Sakai-Kato, N. Matsumoto, T. Toyo’oka, Anal. Chem. 74 (2002) 1915. [15] M.C. Breadmore, S. Shrinivasan, K.A. Wolfe, M.E. Power, J.P. Ferrance, B. Hosticka, P.M. Norris, J.P. Landers, Electrophoresis 23 (2002) 3487.
[16] P. Hatsis, C.A. Lucy, Anal. Chem. 75 (2003) 995. [17] D. Allen, Z. El Rassi, Electrophoresis 24 (2003) 3962. [18] M.C. Breadmore, S. Shrinivasan, J. Karlinsey, J.P. Ferrance, P.M. Norris, J.P. Landers, Electrophoresis 24 (2003) 1261. [19] M. Breadmore, M. Macka, N. Avdalovic, P. Haddad, Analyst 125 (2000) 1235. [20] M. Breadmore, M. Boyce, M. Macka, N. Avdalovic, P. Haddad, Analyst 125 (2000) 799. [21] M. Breadmore, M. Macka, N. Avdalovic, P. Haddad, Anal. Chem. 73 (2001) 820. [22] M. Breadmore, A. Palmer, M. Curran, M. Macka, N. Avdalovic, P. Haddad, Anal. Chem. 74 (2002) 2112. [23] J.P. Hutchinson, M. Macka, N. Avdalovic, P.R. Haddad, J. Chromatogr. A 1039 (2004) 187. [24] N. Ishizuka, H. Kobayashi, H. Minakuchi, K. Nakanishi, K. Hirao, K. Hosoya, T. Ikegami, N. Tanaka, J. Chromatogr. A 960 (2002) 85. [25] Z. Liu, R.A. Wu, H. Zou, Electrophoresis 23 (2002) 3954. [26] M. Breadmore, M. Boyce, M. Macka, N. Avdalovic, P. Haddad, J. Chromatogr. A 892 (2000) 303. [27] J.P. Hutchinson, P. Zakaria, A.R. Bowie, M. Macka, N. Avdalovic, P.R. Haddad, Anal. Chem. 77 (2005) 407.
Preparation and characterisation of anion-exchange latex-coated silica monoliths for capillary electrochromatography Joseph P. Hutchinson a , Emily F. Hilder a , Miroslav Macka a , Nebojsa Avdalovic b , Paul R. Haddad a,∗ a
Australian Centre for Research on Separation Science (ACROSS), School of Chemistry, Faculty of Science, Engineering and Technology, University of Tasmania, Private Bag 75, Hobart, Tasmania 7001, Australia b Dionex Corporation, 1228 Titan Way, Sunnyvale, CA 94088-3603, USA Available online 8 September 2005
Abstract Silica monoliths coated with functionalised latex particles have been prepared for use in monolithic ion-exchange capillary electrochromatography (IE-CEC) for the separation of inorganic anions. The ion-exchange monoliths were prepared using 70 nm quaternary ammonium, anion-exchange latex particles, which were bound electrostatically to a monolithic silica skeleton synthesised in a fused silica capillary. The resulting stationary phases were characterised in terms of their chromatographic performance and capacity. The capacity of a 50 m diameter 25 cm latex-coated silica monolith was found to be 0.342 nanoequivalents and 80,000 theoretical plates per column were typically achieved for weakly retained anions, with lower efficiency being observed for analytes exhibiting strong ion-exchange interaction with the stationary phase. The electroosmotic flow (EOF) was reversed after the latex-coating was applied (−25.96 m2 V−1 s−1 , relative standard deviation (RSD) 2.8%) and resulted in anions being separated in the co-EOF mode. Ion-exchange interactions between the analytes and the stationary phase were manipulated by varying the ionexchange selectivity coefficient and the concentration of a competing ion (phosphate or perchlorate) present in the electrolyte. Large concentrations of competing ion (greater than 1 M phosphate or 200 mM perchlorate) were required to completely suppress ion-exchange interactions, which highlighted the significant retention effects that could be achieved using monolithic columns compared to open tubular columns, without the problems associated with particle-packed columns. The latex-coated silica monoliths were easily produced in bulk quantities and performed reproducibly in acidic electrolytes. The high permeability and beneficial phase ratio makes these columns ideal for micro-LC and preconcentration applications. © 2005 Elsevier B.V. All rights reserved. Keywords: Monoliths; Silica; Latex anion-exchangers; Capillary electrochromatography; Preconcentration
1. Introduction Monolithic stationary phases are an attractive choice for capillary electrochromatography (CEC). The in situ formation of a continuous stationary phase bed having a network of connected through-pores provides an efficient chromatographic medium with high permeability and favourable mass transfer characteristics. Furthermore, monolithic columns can be expected to show considerably higher ion-exchange capacities than open tubular columns due to their larger surface area. Monolithic columns also offer advantages over particle-packed columns. The monolith is formed in situ as a continuous bed within the capillary and
∗
Corresponding author. Tel.: +61 3 6226 2179; fax: +61 3 6226 2858. E-mail address: [email protected] (P.R. Haddad).
0021-9673/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2005.08.076
can be covalently bound to the capillary wall. This eliminates the problems associated with frits such as frit failure and bubble formation, which have often been encountered using particlepacked columns. Monolithic columns have also been shown to be much more permeable than packed capillaries due to the large through-pores formed [1,2] and this increase in permeability leads to shorter flush times, reduced back-pressure and frequently also to enhanced chromatographic performance of the column. The first uniform porous silica monolithic columns for HPLC were reported by Tanaka and co-workers [3] in 1996. In 2000, the sol–gel method was adapted to create silica monoliths in the capillary format and these were derivatised for reversedphase CEC applications [4,5]. In this sol–gel process, chloroand methoxysilanes undergo hydrolytic polymerisation in the presence of aqueous acid and poly(ethylene glycol) (PEG). The
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
silica network must bond to the silanol groups on the capillary wall to prevent shrinkage of the monolithic structure. If shrinkage occurs, voids between the monolithic structure and the capillary wall lead to reduced chromatographic efficiency or the monolithic material may be flushed from the column altogether. The macroporous morphology is formed when transitional structures during spinodal decomposition are fixed permanently by the sol–gel transition. Typically, the silica skeleton formed is 2 m thick and through-pores as large as 10 m can be obtained. Silica monolithic structures differ to those produced from organic polymers by virtue of their bimodal pore distribution and the presence of mesopores within the silica structure. These mesopores typically range from 1 to 50 nm and their size can be controlled by treatment with a base, such as ammonia, after formation of the macroporous silica skeleton. After fabrication, monolithic silica capillary columns have been typically derivatised using octadecyldimethyl-N,Ndiethylaminosilane to produce columns for reversed-phase applications [4,6]. Recently, Allen and El Rassi have investigated a variety of alternative derivatisation procedures for the separation of analytes using capillary electrochromatography. These included hydrophillic silica monolithic stationary phases bearing surface-bound cyano groups for the separation of polar species [7] and amphiphillic silica monoliths possessing surfacebound cationic octadecyl moieties [8] for the separation of neutral and charged species. Other advances in silica monolithic columns prepared for CEC include those with embedded particles [9,10], photochemical polymerisation [11], cationic sol–gel precursors to reverse the EOF [12], derivatisation with chiral selectors [13], sol–gel encapsulation of chiral proteins [14] and monoliths coated with ion-exchange materials [15,16]. Silicabased monoliths for CEC have recently been reviewed in detail by Allen and El Rassi [17]. To date, anion-exchange CEC using monolithic silica columns has received very little attention. The only documented examples in the literature are by Breadmore et al. In the first example [15], monolithic silica capillary columns were fabricated and coated with the soluble cationic polymer poly(diallyldimethylammonium chloride) (PDDAC). The resulting columns were used for the separation of inorganic anions. More recent work by the same authors was performed using dextran sulfate in between multiple PDDAC coatings to form polyelectrolyte multilayers on the monolithic silica surface. The subsequent columns were used for the separation of peptides and inorganic anions [18]. In the present study, monolithic silica columns were prepared in 50 and 75 m I.D. fused-silica capillaries and then coated with latex nanoparticles functionalised with quaternary ammonium groups (AS5A, Dionex). The positively charged latex particles adhere electrostatically to the negatively charged silanol groups on the silica monolith and create an anionexchange stationary phase suitable for ion-exchange capillary electrochromatography (IE-CEC). This approach has been used previously in our group where latexes have been coated onto the wall of fused-silica capillaries and used for open-tubular CEC and preconcentration techniques [19–22]. The stability
11
of the AS5A-coated monolith was tested and the reproducibility and chromatographic performance of the coated monoliths were assessed. This work is part of an on-going investigation to fabricate novel ion-exchange stationary phases to perform separations in CEC and also micro-ion chromatography formats. The ion-exchange selectivity of high capacity columns can be manipulated for faster and more efficient separations and will allow the determination of analytes in real samples, which cannot be analysed without preconcentration methods being performed.
2. Experimental 2.1. Instrumentation and physical characterisation of monoliths The capillary electrophoresis instrument used was an Agilent 3D CE (Waldbronn, Germany). Separations were carried out using 50–100 m I.D. fused-silica capillaries (Polymicro Technologies, Phoenix, AZ, USA) containing a monolithic silica bed with a length of 25 cm, unless otherwise stated. A 75 m I.D. 9.6 cm open-tubular AS5A-coated capillary was attached to the capillary containing the monolith using a 1 cm PTFE sleeve as described previously [23]. A detection segment was formed in this section of capillary by burning a window 1.1 cm away from the end of the monolith and 8.5 from the anode. This allowed UV detection to occur without problems of light scattering on the monolith, which would otherwise be a problem if detection through the monolithic structure was attempted. Anions were separated in the co-electroosmotic flow (co-EOF) mode where both the analyte anions and the EOF migrated towards the anode. The EOF was measured using acetone as a marker. The voltage used for the separations was varied between −1 kV and −30 kV, depending on the concentration of the background electrolyte (BGE) and the temperature was maintained at 25 ◦ C. Injection of anions was performed by hydrodynamic injection at the cathodic end. Scanning electron microscopy (SEM) images were obtained using an ElectroScan ESEM2020 Environmental Scanning Electron Microscope (Wilmington, MA, USA) operated in high vacuum mode using an accelerating voltage of 15 kV. Secondary electrons were detected using an Everhard–Thornley scintillator-type electron detector. The monolithic capillary was cut into 1 cm fragments that were mounted perpendicularly to a 12 mm pin-type aluminium stub using epoxy resin. Highresolution images were obtained by coating the capillary with gold (nominally 40 nm thick). In parallel with synthesis of the monolith in the capillary, synthesis was also carried out in a glass vial of larger volume in order to obtain a sufficient amount of the monolith for the determination of the porous properties. The pore volume and pore size distribution were determined using an Autopore III 9400 mercury intrusion porosimeter (Micromeritics, Norcross, GA, USA). The surface area was calculated from the BET isotherms of nitrogen adsorption and desorption using an ASAP 2010 (Micromeritics).
12
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
2.2. Chemicals AS5A latex (quaternary ammonium functionalised) particles were supplied as an 11% w/v suspension by Dionex (Sunnyvale, CA, USA). Standard solutions of 10 mM Br− , I− (Sigma–Aldrich, Milwaukee, WI, USA), Cl− (May and Baker, Footscray, Australia), BrO3 − , IO3 − , SCN− , CrO4 2− (Ajax, Melbourne, Australia) NO3 − and S2 O3 2− (BDH, Kilsyth, Australia) were prepared from sodium or potassium salts of analytical reagent grade. All samples were prepared in water obtained from a Milli-Q (Millipore, Bedford, MA USA) water purification system. Stock solutions of 100 mM 2-(Nmorpholino)ethanesulfonic acid (MES) and histidine (His) were prepared by dissolution in Milli-Q water of analytical grade reagents purchased from Aldrich. These were combined to form the 50 mM MES/His buffer. Phosphate at pH 7.0 and borate at pH 9.3 were diluted from 50 mM solutions purchased commercially (Agilent, Waldbronn, Germany). All BGEs used were filtered through 0.2 m filters (Millipore) and degassed under vacuum prior to use. 2.3. Procedure for creating latex monoliths The monolithic silica columns were prepared using the method outlined by Ishizuka [24]. Two milliliters of 99% purity tetramethoxysilane was added to 0.44 g PEG (MW = 10 000) and 0.45 g urea in 5 mL 0.01 M acetic acid. All sol–gel reagents were purchased from Sigma–Aldrich (Milwaulkee, WI, USA). The reagents were stirred at 0 ◦ C until a homogenous mixture formed (40 min) and this was then injected into the capillary using highpressure on the Agilent CE instrument and maintained at 40 ◦ C overnight. The capillary was then placed in an oven for 3 h at 120 ◦ C to complete the formation of the mesopores. The monolith was flushed with water and methanol and placed back in the oven at 330 ◦ C for 25 h to remove any residual organic material within the capillary. Finally, the monolithic silica column was flushed with water and methanol prior to latex coating. The monolithic silica columns were then treated with a tenfold aqueous dilution of the AS5A latex suspension provided by Dionex, which had been filtered through a 0.45 m Gelman nylon filter. Ten column volumes of the diluted AS5A latex suspension were flushed through the monolith and the effluent monitored using the UV detector on the instrument. Breakthrough of the latex was evident as a sharp increase in absorbance and was used to indicate that all available sites on the monolith had been saturated with latex. The column was then flushed with water and finally with at least 10 column volumes of BGE prior to use. Between runs the column was flushed with BGE prior to injection. 3. Results and discussion 3.1. Characterisation of latex-coated silica monolithic capillary columns To assess whether the AS5A latex particles had adhered to the silica surface of the monolith, the EOF prior to and after
Table 1 Comparison of the EOF in AS5A-coated open-tubular capillary columns, AS5Acoated monolithic silica capillary columns, and uncoated monolithic silica capillary columns EOF mobility (×10−9 m2 V−1 s−1 ) 50 m I.D. AS5A-coated monolith 1 50 m I.D. AS5A-coated monolith 2 75 m I.D. AS5A-coated monolith 50 m I.D. uncoated monolith 75 m I.D. uncoated monolith 75 m I.D. AS5A-coated OT capillary
−21 ± 2 (n = 11) −26.0 ± 0.1 (n = 13) −33.9 ± 0.3 (n = 3) 29.0 (n = 1) 33.4 (n = 1) −28.7 ± 0.6 (n = 3)
Conditions: columns: all columns were 33.5 cm in length with detection 8.5 cm from the end. BGE: 10 mM perchloric acid; voltage: −30 kV; temperature: 25 ◦ C; EOF marker: acetone.
coating with latex was measured. Table 1 shows that the EOF was reversed after coating to a value similar in magnitude to that reported for an open-tubular capillary coated with AS5A latex particles [15]. The EOF is a result of the bulk flow of ions in the diffuse layer formed at the stationary phase surface to maintain charge balance. This layer is called the electrical double layer and has an associated potential difference known as the zeta potential. Breadmore et al. [15] have previously investigated the dependence of the EOF in a silica monolithic column on the ionic strength of the BGE. They found that the EOF in silica monoliths increased with ionic strength, contrary to the behaviour of an empty capillary. This was explained by Breadmore et al. [15] to be due to the limited EOF in the small mesopores present in the silica skeleton at low ionic strength. Under such conditions, the thickness of the electrical double layer was sufficiently great that overlap from opposing sides of the pore occurred. However, when the ionic strength of the BGE was increased, the thickness of the electrical double layer formed at all surfaces was decreased, reducing the overlap of the electrical double layer within the mesopores. This allowed the mesoporous surface area to contribute to the overall EOF of the column. Typically in an empty capillary, increasing the ionic strength of the BGE decreases the zeta potential, which in turn reduces the magnitude of the EOF. The EOF in the latex-coated monoliths created in this work behaved similarly to the open-tubular columns in that it decreased with increasing ionic strength. This may be a result of the mesopores being obscured by the larger latex particles, thereby reducing the mesoporous contribution to the EOF thought to cause the deviation from empty capillary theory in terms of the velocity of the EOF. Upon coating the silica monolith with the AS5A latex particles, the time taken to flush the column with BGE was reproducibly 30% longer than for the uncoated monolith. The adsorbed latex particles clearly reduced the permeability of the column by decreasing the through-pore diameter. In the coating procedure, approximately six column volumes of a 1% latex suspension were needed to fully coat the monolith, indicating a considerable amount of the latex was retained on the column. The reversal of the EOF in the coated column was further evidence that there was very little bare silica remaining after the coating process.
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
13
Fig. 1. SEM pictures of the: (a) uncoated silica monolithic structure and (b) the AS5A-coated silica monolithic structure.
Several monoliths were viewed using scanning electron microscopy. Fig. 1 shows the structure of the 50 m I.D. AS5Acoated silica monolith in comparison to an uncoated silica monolith. The appearance of the latex-coated monolithic surface is not markedly different to that of the uncoated monolith and an SEM capable of greater resolution would be required for visual identification of the latex particles. However, the size and structure of the through-pores and the relatively thin silica skeleton could be seen clearly. The bimodal silica skeleton (approximately 2 m in width as measured from Fig. 1), is susceptible to decomposition at alkaline pH. This is undesirable as the latex coating would be removed and fragments of the silica skeleton may be dislodged, reducing the permeability and altering the integrity of the electrical circuit established during electrophoresis. The effect of alkaline BGEs was investigated by flushing the column with electrolytes at different pH and evaluating the EOF in successive runs. It was found that an EOF of significantly smaller magnitude occurred after approximately 7 runs at alkaline pH using tris(hydroxymethyl)aminomethane (Tris), phosphate or borate buffered electrolytes. However, when acidic electrolytes such as perchloric, acetic or phosphoric acid between pH 1–3 were used, the EOF was not altered significantly and a single column could be used for a period of several months. For this reason, only acidic BGEs were used in the remainder of this work. Measurement of the EOF was deemed to be the best method for determining the reproducibility of column fabrication since small differences in surface charge markedly alter the magnitude of the EOF. The average EOF and relative standard deviation (RSD) over 15 consecutive runs was found to be −25.96 × 10−9 m2 V−1 s−1 and 2.8%, respectively. A 2 m length of monolithic silica was synthesised in 50 m I.D. fusedsilica capillary and consecutive portions were cut and coated with AS5A latex particles in an identical fashion. These columns were used for testing the repeatability of the coating procedure and for comparing the performance between the columns.
The RSD between columns made from the same batch was 8% (conditions: voltage: −10 kV, BGE: 10 mM HClO4 , temperature 25 ◦ C, EOF marker: acetone) and was similar to that found between separate batches (7.4% RSD). Mercury intrusion porosimetry and BET adsorption isotherms were used to assess the physical properties of the silica monoliths used in this work. These techniques were performed on both uncoated and latex-coated bulk polymerisation samples, which were formed under the same thermal conditions as the monolithic capillary columns. The silica monoliths created were of a bimodal pore structure possessing macropores of 2.2 m in diameter and also mesopores within the silica skeleton of 8.3 nm diameter. The resulting material had a porosity of 70.4%. The surface area was increased from 167 to 186 m2 /g upon coating the silica monolith with latex nanoparticles as measured by BET adsorption isotherms. The behaviour of the monolithic columns under hydrodynamic flow conditions was assessed prior to their use in CEC. Hydrodynamic flow is used for the preconditioning of the column and for quantifying injection volumes into the column. The time required for elution of a neutral, hydrophilic (unretained) marker from the 50 m I.D. AS5A-coated monolith was measured at applied pressures ranging between 950 mbar and 12 bar. Elution was very rapid (<50 s) at pressures exceeding 6 bar, indicating that the columns were well suited to the pressure limitations of the Agilent CE instrument used in this work. 3.2. Ion-exchange capacities of the latex-coated monolithic columns The ion-exchange capacities of the AS5A-coated monolithic silica columns were measured by an adsorption/elution method. Solutions of different anions (10 M) were loaded onto the column hydrodynamically until saturation was observed. Water was flushed through the column to remove any residual anions not bound to the latex. Anions bound to the stationary phase were
14
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
Fig. 2. Ion-exchange capacities of two 50 m I.D. AS5A-coated monolith duplicates for a range of anions and a range of AS5A-coated open-tubular columns of varying diameter.
then eluted using 100 mM perchlorate and the elution monitored with the UV detector on the instrument. The normalised peak area was compared to calibration curves constructed for each anion. Fig. 2 compares the capacity of a latex-coated monolithic column with that of a latex-coated open-tubular column where the capacities shown are the average of the values obtained using six different inorganic anions. There was a large variation in capacity between the two monolithic columns evaluated, despite the fact that they were cut from consecutive portions of the same larger monolithic column and were coated in an identical manner. This suggested that the monolithic structure was not homogenous, although both columns had similar flow-rates and it could be seen under an optical microscope that the monolithic structures formed at the ends of each column were similar. This highlights the difficulty in trying to synthesise reproducible columns, even from the same batch. Theoretical calculations of ion-exchange capacity were performed for the latex-coated monolithic columns and the latexcoated open-tubular columns. Data for the capacity and size of the monodisperse latex particles were provided by Dionex (12 equiv./g and 70 nm, respectively) and the volume of latex coated within the column was calculated by assuming a monolayer of nanoparticles adsorbed to the available surface area. The formation of a monolayer of latex particles has previously been shown by scanning electron microscopy in the case of OTcoated columns [19]. To simplify the calculation, each latex particle was assumed to have the shape of a cube. The theoretical capacity of the monolithic columns was calculated on the basis of a monolayer coverage of latex particles over the monolithic surface area experimentally measured using BET adsorption isotherms. Values obtained for the OT columns are included in Fig. 2 and in all cases were of the same order, but somewhat higher, than the measured value. In the case of the latex-coated monolithic silica column the theoretical capacity was calculated to be approximately 250 times larger than an open tubular latex-coated capillary of the same dimensions. In a related study, Liu et al. [25] calculated that an avidin-coated 50 m I.D. silica monolith had a capacity that was 120 times greater than an open-tubular column of the same dimensions. However, the experimental value of the capacity of the latexcoated monolith was approximately 13 times higher than its
Fig. 3. Breakthrough volumes for 1 mM solutions of a range of analytes on a 50 m I.D. 25 cm AS5A-coated monolith.
open-tubular counterpart in one column and 5.5 times for the second column. This suggested strongly that a monolayer of latex particles had not formed at the monolithic surface. This may be attributed to a lower density of surface silanol groups on the monolith or it may be related to the physical morphology of the material, which might impede monolayer adsorption of the spherical latex particles. Breakthrough studies were performed by pumping a solution of the desired analyte through the capillary at various flow-rates and monitoring the UV absorbance of the effluent. Fig. 3 shows the results obtained and confirms that the breakthrough volume increased with increasing ion-exchange selectivity coefficient for the analytes tested. Moreover, it is evident that the magnitude of the flow-rate used had little effect on the breakthrough volumes obtained due to the improved mass transfer characteristics of the monolithic column. The random nature of the through-pores in the monolith increases the tortuosity of flow and improves mass transfer by reducing the distance of diffusion to the stationary phase, thereby aiding effective adsorption of the analytes. This indicates that these columns may be suitable for on-line preconcentration purposes. 3.3. Separation efficiency of the latex-coated monoliths Variation in the separation voltage resulted in changes in the linear velocity of the EOF. Van Deemter plots for a variety of analyte anions in a perchlorate BGE on two AS5A-coated monoliths were constructed and Fig. 4 shows these plots for bromide at different linear velocities. The minimum value of plate height (approximately 25 m) was observed at a linear velocity of about 1 mm/s, which corresponded to an applied voltage of approximately −15 kV. This represented an efficiency of 40,000 theoretical plates/m. This value was influenced by the masstransfer effects due to the existence of ion-exchange interactions between the analyte and the stationary phases, and also the fact that the system used involved the coupling of two capillary segments (i.e. the separation segment and the detection segment). The latter effect has been shown to considerably reduce the efficiency [23]. Peak efficiency was also measured as a function of the ionic strength of the BGE at constant voltage and the results are given in Table 2. A low voltage was required to avoid the effects of Joule heating as the BGE concentration was increased to 200 mM to reduce the ion-exchange interaction of iodide,
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
15
Table 2 Effect of BGE concentration on separation efficiency [ClO4 − ] in BGE (mM)
Theoretical plates per metre Br−
1 5 10 20 40 80 160 200
20,980 20,240 50,650 72,820 77,680 74,900 77,610 36,740
NO3 −
I−
30,820 19,600 50,430 61,020 34,420
1,320 1,820 1,200 3,040
35,980 8,790
27,420 7,840
a
a
SCN−
S2 O3 −
4,720 4,380 6,640 34,670 31,730 13,310 10,740
30,260 61,820 26,520
Conditions: column: 25 cm 75 m I.D. AS5A-coated silica monolith attached to a 10 cm 75 m I.D. open-tubular AS5A-coated column with a window at 8.5 cm via a PTFE sleeve; injection: 40 mbar for 40 s. Voltage: −1 kV; temperature: 25 ◦ C. a NO − and I− were co-eluted with this electrolyte. 3
thiocyanate and thiosulfate. In Table 2 it can be seen that as the ion-exchange interaction is decreased (by increasing the ionic strength of the BGE), the peak efficiency is increased. For example, bromide had an efficiency of approximately 21,000 plates/m when separation was performed using 1 mM perchlorate in the BGE and the efficiency increased to 80,000 plates/m when using 160 mM perchlorate in the BGE. The reason the efficiency was seen to be reduced at 200 mM was due to the reduction of the zeta potential and thus the magnitude of the EOF at high ionic strength.
mide, as indicated by the fact that its effective mobility was still increasing in the negative direction. On the other hand, when perchlorate was used as the competing ion (Fig. 5b) only 40 mM was required to limit the ion-exchange interaction of bromide with the stationary phase during the separation, while higher concentrations were needed to achieve the same goal for analyte ions of higher ion-exchange selectivity, such as thiocyanate. The separation selectivity could be controlled by the nature and concentration of the BGE, but it should be noted that increasing the ionic strength of the BGE increased the current during electrophoresis and necessitated the use of a lower applied voltage
3.4. Variation of electrolyte composition The chromatographic performance of the latex-coated silica monoliths was assessed by altering the composition and concentration of the BGE, predominantly with a view to manipulating the extent of the ion-exchange interactions between the analytes and the stationary phase. Fig. 5 shows the retention effects occurring in phosphate and perchlorate BGEs. In Fig. 5a, the retention of analyte anions having low ion-exchange selectivity coefficients are shown for phosphate BGEs of varying concentration. It can be seen that even at a concentration of phosphate of 1000 mM, ion-exchange effects were still apparent for bro-
Fig. 4. Van Deemter plots prepared using bromide as analyte at a range of applied voltages for two latex-coated silica monolithic columns. Conditions: columns: 25 cm 50 m I.D. AS5A-coated monolith attached via PTFE sleeve to 9.6 cm 75 m I.D. AS5A-coated open-tubular detection segment with window 8.5 cm from end; BGE: 5 mM perchloric acid; Injection: 50 mbar 5 s of 1.4 mM Br− ; temperature: 25 ◦ C.
Fig. 5. Selectivity effects for a AS5A-coated silica monolith using: (a) phosphoric acid BGE and (b) perchloric acid BGE Conditions: column: 50 m I.D. 24.1 cm AS5A-coated silica monolith attached via PTFE sleeve to 9.6 cm 75 m I.D. AS5A-coated open-tubular column with detection window 8.5 cm from end; Injection: 40 mbar 40 s of 1 mM solutions of indicated anions. Voltage: between −10 and −1 kV as required to reduce Joule heating; temperature: 25 ◦ C.
16
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
3.5. Latex-coated monolithic columns for in-line sample preconcentration
Fig. 6. IE-CEC separations of inorganic anions on a latex-coated silica monolith when the concentration of perchlorate as competing anion was varied. Detection was performed on-column through the monolithic bed. Conditions: column: 33.4 cm × 50 m I.D. AS5A-coated silica monolith with detection on-column 8.5 cm from anodic end; injection: 50 mbar 20 s of 2 mM solutions of the indicated anions; BGE: Tris/ClO4 − at varying concentrations; voltage: −15 kV; temperature: 25 ◦ C; UV detection: 195 nm.
to avoid Joule heating effects. This in turn lowered the speed of the separations achieved. The replicate 50 m I.D. AS5A-coated monolith (column 2) was also evaluated for its chromatographic performance and it exhibited the same selectivity effects as column 1, but with slightly lower retention as expected from its lower ion-exchange capacity. It was anticipated that the increase in the phase ratio when using latex-coated monolithic columns compared to latex-coated open-tubular columns would improve separation efficiency. In open-tubular CEC separations it is known that the capillary internal diameter should be less than 25 m to provide optimised efficiency as Taylor dispersion (resistance to mass transport in the mobile phase) increases steeply with column diameter. The monolithic stationary phases used in this work gave better efficiencies than those found in previous studies using AS5A latex-coated open-tubular columns [19,26], which achieved efficiencies of about 10,000 plates per metre. However, as stated earlier there are several reasons to believe that the efficiencies achieved in this present work may be less than optimal due to the coupling of capillaries. For example, the efficiency was found to be much greater when detection was performed through a 1 cm gap in the monolithic bed rather than by coupling a detection capillary segment. This can be formed by injecting a small amount of air into the capillary between the sol–gel solution prior to heat treatment and formation of the monolithic structure within the capillary. Such separations performed in this way using perchlorate BGEs of varying ionic strength are shown in Fig. 6.
In addition to the CEC separations described above, a further potential use of the latex-coated monoliths is for in-line sample preconcentration. Here, a short capillary segment containing the monolith (the preconcentration segment) is coupled to a longer segment of open capillary (the separation segment), in which the separation and detection are performed. This approach is similar to that used in a previous study in which we utilised latex-coated polymeric monoliths for in-line sample preconcentration [27]. It was necessary for the separation capillary to be coated with a positively charged polymer (PDDAC) in order to reverse the EOF so it was in the same direction as that for the preconcentration column. This provided a fast separation of anions in the co-EOF mode and minimised disturbances of the compressed analyte zone as it left the preconcentration segment and entered the separation segment. Fused-silica capillary of 50 m I.D. was found to be the best choice for creating the latex-coated silica monolithic preconcentration columns. Use of such relatively small internal diameter capillaries minimised shrinkage of the monolith from the inner wall of the capillary, which was a problem in larger diameter capillaries. Using smaller diameter preconcentration capillaries also resulted in lower operating currents and reduced the effects of Joule heating. Decoupling the preconcentration segment and the separation segment allowed separate optimisation of the dimensions of each segment. Detection limits can be improved since a larger diameter separation capillary can be used, which improves detection by providing a longer optical path-length. After sample loading, the bound analyte ions were eluted from the preconcentration segment and focused into a compact band using a transient isotachophoretic gradient, as described previously [21]. A fluoride/5-sulfoisophthalate discontinuous electrolyte system was used for this purpose. Upon the application of voltage the preconcentrated anions migrate in front of the 5-sulfoisophthalate strong electrolyte (SE) entering the preconcentration segment and when these ions reach the separation segment, ion-exchange interaction ceases and an electrophoretic separation occurs in the fluoride weak electrolyte (WE) filling the separation capillary. Fluoride has an electrophoretic mobility close to those of the preconcentrated analyte anions and this minimises the effects of electromigrational dispersion and produces satisfactory peak shapes for these analytes. 5-Sulfoisophthalate is an ideal SE as its ion-exchange selectivity coefficient is 14,000 times larger than that of fluoride and it has a suitably low electrophoretic mobility of −33.77 × 10−9 m2 V−1 s−1 , allowing a large range of anions to migrate electrophoretically in front of the isotachophoretic gradient formed. Fig. 7 shows the separation of anions obtained using this method, where the concentration of the fluoride WE was adjusted from 1 mM (which allowed binding of a range of ions on the preconcentration segment) to 100 mM (which permitted only strongly retained anions like chromate and iodide to be preconcentrated). It should be pointed out that part of the preconcentration procedure [21] involved forward-flushing of the monolith with WE after sample loading. Depending on the concentration
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
17
4. Conclusions
Fig. 7. Ion-exchange preconcentration and elution using a transient isotachophoretic gradient where the WE was introduced past the preconcentration zone using: (a) 1 mM F− and (b) 100 mM F− . Conditions: column: 25 cm 50 m I.D. AS5A-coated monolithic silica column coupled via PTFE sleeve to 10 cm 75 m I.D. PDDAC-coated open-tubular column with UV detection at 8.5 cm from end; WE: NaF used at concentrations of (a) 1 mM and (b) 100 mM and introduced from cathodic end for 1.1 column volumes; SE: 5-sulfoisophthalate was used at matching concentration to the WE used; injection: 6 bar 4 min of 1 M Br− , NO3 − , I− , SCN− , IO3 − and BrO3 − in water; voltage: −10 kV; temperature: 25 ◦ C; UV detection: 195 nm.
of the WE, this step can cause some analytes to be flushed to waste. This effect is apparent in Fig. 7b, in which the 100 mM fluoride WE has eluted the weakly retained analytes to waste and they therefore do not appear in the final chromatogram. The recoveries for chromate and iodide in Fig. 7b were 97 and 91%, respectively. A preconcentration factor of approximately 480 was achieved in this study for iodide when 4.8 capillary volumes of the sample were introduced into the preconcentration column compared to a standard CE injection occupying 1% of the column. In comparison, when latex-coated monolithic polymer stationary phases were used, a preconcentration factor of approximately 200 was achieved for the method used to preconcentrate iodide in seawater. However, in the previous study [27], the detection limit was much lower for iodide due to the analyte focusing achieved when using a latex-coated OT separation column coupled to the preconcentration column. The capacity of both the latex-coated polymer (17.1 picoequivalents/cm for a 75 m I.D. column) [27] used previously and the silica monoliths (13.8 picoequivalents/cm for a 50 m I.D. column) used in this study were found to be similar. However, it was possible to synthesise polymer monolithic phases in larger I.D. capillaries (250 m), which increased the volume of stationary phase of the preconcentration column and hence the effective capacity to a value of 386 picoequivalents/cm. The performance of this system could be enhanced through the use of a higher capacity stationary phase, which would be achievable by manipulating the pore size of the monolith or using latex particles of higher ion-exchange capacity.
There is a significant increase in the retention of analytes in CEC using latex-coated monolithic columns compared to their latex-coated open-tubular counterparts. This increase in retention of analytes can be attributed to the higher phase ratio and higher ion-exchange capacity of monolithic columns. However, the monolith stationary phases also required the use high ionic strength electrolytes in order to vary selectivity by manipulating the extent of ion-exchange interactions between the analytes and the stationary phase. This in turn led to problems with Joule heating effects and lower separation efficiencies for highly retained analytes. There were two disadvantages associated with using latexcoated monolithic silica columns. First, the silica backbone imposed a requirement for low pH electrolytes and its fragile nature could lead to blockages and reduced lifetime of the column. Second, there was a restriction in the choice of the capillary internal diameter used to house the monolithic columns because of shrinkage of the monolithic structure away from the inner wall occurring in larger internal diameter capillaries. Preconcentration of a range of inorganic anions was performed on a latex-coated monolithic silica preconcentration segment. Preconcentrated anions were eluted using a transient isotachophoretic gradient and separated electrophoretically when coupled with a separate segment of capillary designed for this purpose. However, the ion-exchange capacity of the preconcentration bed needs to be increased to enable reproducible preconcentration of anions possessing a wide range of IE selectivity coefficients and for the retention of strongly retained anions in high ionic strength matrixes. Latex-coated silica monolithic stationary phases are highly permeable compared to particle-packed columns and are of higher capacity than OT columns. These characteristics make such columns suitable for analyte preconcentration in real samples, which challenge the detection limit of traditional detection systems. Possible applications which will be focused upon in future work using latex-coated monolithic columns include the determination of trace contaminants such as (perchlorate) in drinking water and analysing trace inorganic anions in difficult matrices such as the determination of iodide and thiocyanate in seawater samples. Acknowledgements Financial support from the Australian Research Council is gratefully acknowledged. The authors would also like to thank David Steele of the Central Science Laboratory at the University of Tasmania for the SEM images. References [1] N. Vervoort, P. Gzil, G.V. Baron, G. Desmet, Anal. Chem. 75 (2003) 843. [2] N. Tanaka, H. Kobayashi, N. Ishizuka, H. Minakuchi, K. Nakanishi, K. Hosoya, T. Ikegami, J. Chromatogr. A 965 (2002) 35. [3] H. Minakuchi, K. Nakanishi, N. Soga, N. Ishizuka, N. Tanaka, Anal. Chem. 68 (1996) 3498.
18
J.P. Hutchinson et al. / J. Chromatogr. A 1109 (2006) 10–18
[4] N. Ishizuka, H. Minakuchi, K. Nakanishi, N. Soga, H. Nagayama, K. Hosoya, N. Tanaka, Anal. Chem. 72 (2000) 1275. [5] N. Ishizuka, K. Nakanishi, K. Hirao, J. Sol–Gel Sci. Technol. 19 (2000) 371. [6] N. Tanaka, H. Nagayama, H. Kobayashi, T. Ikegami, K. Hosoya, N. Ishizuka, H. Minakuchi, K. Nakanishi, K. Cabrera, D. Lubda, J. High Resolut. Chromatogr. 23 (2000) 111. [7] D. Allen, Z. El Rassi, J. Chromatogr. A 1029 (2004) 239. [8] D. Allen, Z. El Rassi, Analyst 128 (2003) 1249. [9] M.T. Dulay, R.P. Kulkarni, R.N. Zare, Anal. Chem. 70 (1998) 5103. [10] C.K. Ratnayake, C.S. Oh, M.P. Henry, J. High Resolut. Chromatogr. 23 (2000) 81. [11] M.T. Dulay, J.P. Quirino, B.D. Bennett, M. Kato, R.N. Zare, Anal. Chem. 73 (2001) 3921. [12] J.D. Hayes, A. Malik, Anal. Chem. 72 (2000) 4090. [13] J. Kang, D. Wistuba, V. Schurig, Electrophoresis 23 (2002) 1116. [14] M. Kato, K. Sakai-Kato, N. Matsumoto, T. Toyo’oka, Anal. Chem. 74 (2002) 1915. [15] M.C. Breadmore, S. Shrinivasan, K.A. Wolfe, M.E. Power, J.P. Ferrance, B. Hosticka, P.M. Norris, J.P. Landers, Electrophoresis 23 (2002) 3487.
[16] P. Hatsis, C.A. Lucy, Anal. Chem. 75 (2003) 995. [17] D. Allen, Z. El Rassi, Electrophoresis 24 (2003) 3962. [18] M.C. Breadmore, S. Shrinivasan, J. Karlinsey, J.P. Ferrance, P.M. Norris, J.P. Landers, Electrophoresis 24 (2003) 1261. [19] M. Breadmore, M. Macka, N. Avdalovic, P. Haddad, Analyst 125 (2000) 1235. [20] M. Breadmore, M. Boyce, M. Macka, N. Avdalovic, P. Haddad, Analyst 125 (2000) 799. [21] M. Breadmore, M. Macka, N. Avdalovic, P. Haddad, Anal. Chem. 73 (2001) 820. [22] M. Breadmore, A. Palmer, M. Curran, M. Macka, N. Avdalovic, P. Haddad, Anal. Chem. 74 (2002) 2112. [23] J.P. Hutchinson, M. Macka, N. Avdalovic, P.R. Haddad, J. Chromatogr. A 1039 (2004) 187. [24] N. Ishizuka, H. Kobayashi, H. Minakuchi, K. Nakanishi, K. Hirao, K. Hosoya, T. Ikegami, N. Tanaka, J. Chromatogr. A 960 (2002) 85. [25] Z. Liu, R.A. Wu, H. Zou, Electrophoresis 23 (2002) 3954. [26] M. Breadmore, M. Boyce, M. Macka, N. Avdalovic, P. Haddad, J. Chromatogr. A 892 (2000) 303. [27] J.P. Hutchinson, P. Zakaria, A.R. Bowie, M. Macka, N. Avdalovic, P.R. Haddad, Anal. Chem. 77 (2005) 407.