Chemiluminescence in Analytical Chemistry

concentration may not be too high or else the baseline noise produced by the CL background becomes unstable), the optimum concentration ratio of Co2 to HIBA was found to be 8.10 3 M. La3 , Ce4 , Pr3 , and Nd3 were successfully separated and detected providing detection limits of 33, 27, 42, and 50 fmol, respectively.

Zhang et al. have first used online indirect detection for analysis of catecholamines (CAS) and catechol (CAT) [90] using the CE-CL system reported above [88]. In this case, the high and constant CL background is obtained by the CL reaction of luminol enhanced with Co(II). The analytes complex with Co(II) and reduce the free Co(II) concentration, and thus the CL intensity decreases. The degree of CL suppression is a measure of the analyte concentration. The authors propose a new mixing mode of the analytes with the CL reagent in which luminol is used as a component of the electrophoretic carrier; H2O2 and Co(II) are introduced by postcapillary. In this way, luminol, H2O2, and Co(II) meet just at the detection window simultaneously. Against the mixing mode previously reported by Liao et al. [86], who used luminol and H2O2 as electrophoretic carrier and catalyst solution of a carbonate buffer containing copper sulfate, the formation of bubbles produced by H2O2 in the presence of base is impeded, which prevents the electrophoretic current and CL background from being unsteady, an increase in the noise, and a decrease in the separation efficiency. Sodium dodecyl sulfate was used in the separation of CAT, epinephrine, norepinephrine, and dopamine, obtaining detection limits of 87, 51, 22, and 38 fmol, respectively. Six amino acids such as Arg, Hyp, Lys, His, Glu, and Asp were detected, yielding better limits of detection than the ones reported by Liao et al. in their first study.

Tsukagoshi’s group developed a new CE apparatus with an online CL detection using the luminol-H2O2 system for analyzing heme proteins [91]. It was found that iron (III), sulfate, hematin as an iron (III) porphyrin complex, and various heme proteins migrated and could be detected. They used the same threeway joint for mixing a CL reagent solution with an eluate from the capillary and the cell structure for detecting a low CL signal previously described [70, 74]. However, treatment of the CL solution (luminol H2O2) was considered due to the drastic change of the CL intensity of the solution upon standing. In this case, the CL intensity quickly decreased and became about one-tenth of that of the initial solution within 1 h; an almost constant CL intensity was observed after about 12 h. For this reason, a CL reagent solution after being left for more than one night was used in this study because a fresh CL reagent solution provided high baseline noise levels and hence no reproducible results. They used a fresh capillary tube (50 m id, 70 cm length) treated with 1 mol/L sodium hydroxide for 30 min and washed with distilled water and a migration buffer solution 10 mmol/L carbonate, pH 10. After a capillary tube was filled with the buffer solu-

tion in advance, a CL reagent solution (5 mmol/L luminol and 25 mmol/L H2O2 aqueous solution) was fed at a rate of 0.6 L/min by a pump until a definitive

electric current was obtained. Then, only the high voltage was removed, and a

sample solution prepared by the carbonate buffer was introduced into a capillary tube having a positive electrode side for 10 s from a height of 30 cm by siphoning. After introduction of the sample solution, a voltage of 0–20 kV was gradually applied for 60 s. Monitoring was started just after the voltage reached 20 kV.

Though CE has shown excellent performances so far for the separation of many compounds having various molecular weights, it is not always satisfactory for the separation of biopolymers, such as proteins, glycoproteins, and lipoproteins due to the adsorption of proteins onto the inner wall of a capillary tube (and the low sensitivity in the detection of protein), which promote the absorption phenomenon due to the high concentration of the protein sample. However, in this case considerable sharp and symmetrical peaks were observed for all protein samples, in spite of turbulent mixing of the analyte with a CL solution at the end of the separation capillary. These satisfactory results are due to several reasons: first, since all protein samples were migrated at pH 10, which is higher than or equal to the isoelectric point values of the protein, the interaction between protein surfaces and negative charges due to silanol groups on the inner wall of capillary must be either very small or negligible, and second, the concentration of protein samples, which is much lower than that used for ordinary spectrophotometric and fluorimetric detection, was subjected to the CE-CL method. This CE-CL method was about 104 times as sensitive as the conventional CE-absorption detection system for the detection of hemoglobin.

3.4Chemiluminescence Reaction with Tris(2,2′-bipyridine)ruthenium (II)

In the ruthenium tris-bipyridine system, an orange emission at 610 nm arises when the excited stated [Ru(bpy)32 ] decays to the ground state. Ru(bpy)32 is the stable species in the solution and the reactive species—Ru(bpy)33 —can be generated from Ru(bpy)32 on the electrode surface by oxidation at about 1.3 V. Adding Ru(bpy)32 to the electrolyte and using an end-column electrode to convert the Ru(bpy)32 into the active Ru(bpy)33 form allow a simple and sensitive ECL detection mode. The reaction lends itself to electrochemical control due to the electrochemically induced interconversion of the key oxidation states:

Ru(bpy)32 → Ru(bpy)33 e

Ru(bpy)33 reductant → product [Ru(bpy)32 ]* [Ru(bpy)32 ]* → Ru(bpy)32 hν

Advantages include: in this CL system the reagent is regenerated and it can be recycled, and derivatization is not required for many classes of compounds. Many aliphatic amines such as alkylamines, amino acids, proteins, antibiotics such as erythromycin and clindamycin, and NADH, among others can par-

ticipate in this reaction, which is compatible with FIA and HPLC solvent systems [92]. ECL employing Ru(bpy)32 offers an attractive detection scheme for CE because of the solubility and stability of the reagents in aqueous media [93]. Moreover, the efficiency of Ru(bpy)32 ECL is high over a wide pH range of pH making it compatible with most buffers systems commonly used in CE.

In 1997, Nieman’s group used this system for the first time in CE introducing inactive Ru(bpy)32 into the electrophoretic buffer and generating electrochemically, online the active Ru(bpy)33 species just inside the outlet of the capillary [94]. This was done by applying 1.25 V versus Ag/AgCl to a Pt wire inserted 3 mm into the outlet of the capillary. In this way, analyte bands exiting the separation capillary react with the Ru(bpy)33 and produce light. The outlet is placed within a parabolic mirror that directs the emitted photons to a photon counting PMT. Concentration of Ru(bpy)32 in the electrolyte is an important parameter to be optimized because of its large impact on the background CL signal and dynamic range. The performance of the system was demonstrated using a series of β-blockers, a class of amine compounds that block the effect of norepinephrine on the adrenergic receptors. Detection limit for oxprenolol was 0.6 µg/mL with a separation efficiency of 15,000 plates. In the same period, Tsukasoshi et al. [95] found that the emetine dithiocarbamate Cu(II) complex, prepared from the emetine alkaloid, carbon disulfide, and Cu(II), showed a sensitive response on a Ru(bpy)32 ECL system, developing a CE-CL detection method for the analysis of emetine. They used an outline CE apparatus with an ECL detector. Ru(bpy)32 was fed at a rate of 40 mL/min by a pump, being oxidized using an electrolytic current of 100 mA in an electrochemical reactor and then mixed with the eluate at the tip of the capillary tube. A sample solution was introduced into the capillary tube having a positive electrode side for 20 min from 15-cm height by siphoning and a voltage of 0–20 kV was gradually applied for 60 s. A photon counter measured the CL signal at the tip of capillary. A combination of the dithiocarbamate complex formation of transition metal ions and their CL response to Ru(bpy)32 ECL is expected to be useful for the analysis of transition metal ions.

Another simple apparatus for postcolumn ECL detection [96] employs a conductive joint to isolate the separation field from the potential need to drive the ECL due to the electric currents generated in capillaries with id’s greater than 25 µm, which greatly affects the faradaic currents at microelectrodes. This is accomplished by constructing a porous joint in the separation capillary (100 µm id, 360 µm od, 60-cm length). A 3-mm segment of the polyimide coating of the fused silica capillary, 5 cm from the capillary’s end, is removed with hot concentrated sulfuric acid and the capillary is rigidly fixed to a Plexiglas plate, which serves to reinforce the capillary at the position where the polyimide coating has been removed. Teflon tubing provides a contact surface between the capillary and the Plexiglas plate. Next the entire capillary is placed in the cap of a Nalgene

bottle, which later serves as a reservoir for the ground end of the separation capillary. The ends of the capillary are passed through the holes drilled in the Nalgene cap, which are then filled with epoxy. A porous joint is then etched in the capillary (where the polyimide coating has been removed) by immersing the capillary in 40% BF for 1–2 h. Capillaries are then filled with the running buffer and a potential of 10 kV is applied across the capillary. The ECL cell that houses the capillary, ECL reagent, Pt working, Pt auxiliary, and Ag/AgCl reference electrodes are fabricated from a Nalgene bottle cap and filled with 1 nM Ru(bpy)32 and Na2HPO4 (pH 9). The ECL signal is detected by a PMT positioned directly above the working electrode. The device is shown in Figure 9, Chapter 9 of this book. In this case, postcolumn and precolumn reagent addition were compared, showing that although the addition of the CL reagent to the running buffer and on-column detection ideally would lead to higher efficiency separations than the postcolumn arrangement, differences in migration between analytes and Ru(bpy)32 lead to zone broadening. Moreover the absorption of Ru(bpy)32 onto the silica walls of the separation capillary with an equilibration time of several hours and the perturbation in equilibrium produced when the capillary is flushed with dilute NaOH, water and finally buffer, hinder the use of on-column detection, suggesting the postcolumn addition of CL reagent to overcome these problems.

In this sense, a postcapillary reservoir of Ru(bpy)32 has been used for in situ generation of Ru(bpy)33 [97]. Ru(bpy)32 is added postcapillary as a small reservoir ( 100 L) at the interface of the separation capillary and the detection electrochemical cell and is then converted to Ru(bpy)33 at a carbon microfiber for reaction with eluting amines or amino acids. This detection approach has been found to provide a reproducible electrophoretic separation compatible with the nanoliter detection volumes required to maintain CE separation efficiencies, the major advantage being that the electrophoresis will not be inhibited by the presence of Ru(bpy)32 in the running buffer. Figure 13 shows the CE separation of triethylamine (TEA), proline, valine, and serine at pH 9.5 using CL detection. Detection limits range from approximately 100 nM for TEA and proline, the most efficient luminescent species, to approximately 100 M for serine.

One problem associated with this design is that the Ru(bpy)32 reservoir evaporates over time and the Ru(bpy)33 concentration changes as the CE capillary effluent dilutes it, which affect both sensitivity and reproducibility of the CL response. To overcome this problem, recently a new in situ–generated Ru(bpy)33 CL cell has been proposed [98]. In this design, Ru(bpy)32 is continuously delivered to the cell and Ru(bpy)33 is then generated at the interface of the separation capillary and the working electrode. Electrochemical control of the production of Ru(bpy)33 at the distal end of the separation capillary without interference from the CE current is provided and finally the ECL process is cou-

Figure 13 Electropherogram of selected amino acids with end-column addition of 1 mM Ru (bpy)32 . Separation conditions: 20 kV with injection of analytes for 8 s at 20 kV. Capillary, 75 m id, 62 cm long with a 4-cm detection capillary. Buffer 15 mM borate, pH 9.5. The electrode used for in situ generation of Ru(bpy)33 was a 35-m-diameter carbon fiber, 3 mm long held at 1.15 V versus a saturated calomel electrode. The PMT was biased at 900 V. Peak identification: (1) 100 fmol TEA, (2) 70 fmol proline; (3) 1.6 pmol valine, (4) 50 pmol serine.* Injection points. (From Ref. 97, with permission.)

pled to an optical system for monitoring light emission. The authors used a CE homemade apparatus to perform electrophoretic separation and electrokinetic injections. A 2–3-mm detection window at the end of the separation capillary was formed by thermally removing the polyimide coating and this separation capillary was inserted into the reaction tube, Ru(bpy)33 reagent was delivered by a syringe pump to the reaction cell at a flow rate of 10 L/min. This detection cell is schematically represented in Figure 14. For precise electrochemical control of the conversion of Ru(bpy)32 to Ru(bpy)33 , the CE current is isolated from the detection capillary using an on-column fracture. The efficacy of this approach was shown in the detection of different amino acids. Further work is expected on the development of end-column detection, noise reduction strategies involving more sensitive PMTs cooled to minimize dark noise, the use of other capillary detection modes, and the application to a wide range of amino acids including derivatives that may be produced as a consequence of protein sequence analysis.

Figure 14 Schematic representation of the Ru(bpy)33 CL in situ detection cell. (From Ref. 98, with permission.)

3.5Chemiluminescence Reaction with Potassium Permanganate in Acidic Medium

Oxidation of catecholamines by potassium permanganate in an acidic medium is known to produce CL, being used for the first time in 1997 in CE [99]. During the preliminary investigation for the analysis of catecholamines, the conventional end-column detection mode used in previous work [87] was first attempted by the authors. The end of a separation capillary was directly inserted into a short piece of a fused-silica reaction/detection capillary, and a tee connector was used to join the separation capillary, oxidant capillary, and reaction/detection capillary together. Acidic permanganate solution was fed through the oxidant capillary and mixed with analytes at the end of the separation capillary, which was inside the reaction/detection capillary. The downstream end of the reaction/detection capillary was dipped in a grounded buffer reservoir to complete the CE electric circuit. A 5-mm window was formed on the reaction/detection capillary (starting at the point where the inner separation capillary terminated) by burning off the polyimide coating. However, CL emission of catecholamines was not observed and the colorless buffer solution in the anodic reservoir gradually turned pink with successive CE experiments. This occurs because the oxidizing agent (permanganate anion) added at the column end migrates electrophoretically toward the anode

and preoxidizes the analytes inside the capillary; therefore, no CL reaction occurs at the column end. To prevent the permanganate from the backstream migrating into the capillary, the reaction/detection zone at the column end was separated from the high-voltage electric field, developing for the first time an off-column CL detection device in CE. A porous polymer joint is easily constructed by fracturing the capillary followed by covering the fracture with a thin layer of cellulose acetate membrane. This porous joint, rather than the end of the capillary, was submerged in a buffer reservoir along the cathode (Fig. 15). The applied voltage was dropped across the capillary prior to the porous joint and the resulting EOF acted as a pump to push the analytes through the short section of capillary after the joint. The analytes mixed with permanganate emitted CL in a field-free region at the column outlet. The feasibility of this off-column CL detection mode was demonstrated in the CE of serotonin, catecholamines, and catechol. This detection mode is useful in cases where the CL reagent or catalyst added at the column end must not stream back into the separation capillary and degradation and decomposition of analytes may occur inside the capillary before they reach the column end. However, the only limitation is that at least some EOF is needed to push the analytes past the grounded joint. Further research is expected about the optimization of this system to enhance the sensitivity and efficiency.

Figure 15 Schematic diagram of the CE with off-column CL detection system. (From Ref. 99, with permission.)

3.6 Chemiluminescence Reaction with Firefly Luciferase

Recent applications in the field of biochemical analysis have been developed based on the highly efficient firefly luciferin-luciferase reaction, a bioluminescence reaction in which two steps can be considered:

Luciferin ATP → Adenyl-luciferin PPi

Adenyl-luciferin O2 → Oxyluciferin AMP CO2 light

the emission maximum occurring at 562 nm.

In this reaction, the most important analyte is adenosine 5′-triphosphate (ATP), appearing either directly or coupled with other enzymatic systems involving ATP as a reactant or product.

With the newly proposed detector, Dadoo et al. [84] adapted this bioluminescence reaction to determine ATP. A selective and sensitive determination is achieved because the use of CE as a separation technique minimizes the effect of several interfering substances such as some anions (e.g., SCN , I ) that inhibit the reaction decreasing the luminescence emission, and even some nucleotides that generate light in this reaction but with lower intensity. A detection limit of 5 nM, approximately 3 orders of magnitude lower than using UV detection, was obtained.

The same reaction was recently proposed to detect creatine kinase (CK), an enzyme of high clinical significance in relation to the investigation of skeletal muscle disease and the diagnosis of myocardial infarct or cerebrovascular accidents. As ATP is a reaction product obtained from the reaction of ADP with creatine phosphate catalyzed by CK, this enzyme can be indirectly measured by the CL intensity read from the subsequent reaction of ATP with luciferin. Using the technique of electrophoretically mediated microanalysis (EMMA), it is possible to detect the enzyme using nanoliter volumes of biological sample with an improved speed and simplicity with respect to a conventional colorimetric method [100].

By application of EMMA Regehr and Regnier developed several assays for enzymes that produce (galactose oxidase and glucose oxidase) or consume (catalase) hydrogen peroxide. Unlabeled enzymes were determined in the femtomole mass range, while detection limits of less than 10,000 molecules were reported for catalase [101].

4. RECENT ADVANCES AND FUTURE PERSPECTIVES

4.1 Micromachining Techniques

Recent trends are focused on the use of micromaching techniques to form miniaturized capillary geometries in planar microdevices. These superminiaturized sys-

tems are being considered to overcome the problem of unsatisfactory detection limits characteristic of standard CE setups. Although in the last five years an important number of publications have been reported in this area, several devices are still being perfected offering great promise for the rapid and efficient processing of various types of samples, fundamentally in biological and clinical assays. A recent trend in miniaturization is the application of micromachining techniques (photolithography and chemical etching) in the fabrication of a complex manifold of flow channels on a microchip, capable of sample injection, pretreatment, and separation [102, 103]. The nature of electrokinetically driven systems as shown by CE makes it suitable for integration on a planar device, giving highly efficient separations in short capillaries together with a considerable reduction of analysis time and not requiring high-pressure pumps or gas supply. As electroosmotic flow velocity and electrophoretic migration depend only on the strength of the applied field, the separation efficiency is exclusively related to the voltage installed across the separation capillary and not to its length. Several materials such as planar glass, fused silica wafers, and quartz have been used to construct devices with different geometry and sizes. Since the first contribution by Manz’s group in 1991 [104], introducing the concept of the CETAS system (capillary electrophoresis micro-total analysis system), further advances have been achieved and revised, reducing considerably the microchip size and extending the field of applications [105, 106].

Recently CL detection based on the horseradish peroxidase (HRP)-cata- lyzed reaction of luminol with peroxide has been investigated as a postseparation detection scheme for microchip-based CE [107]. Evaluation of CL detection on microchips was performed using the luminol reaction with various forms of HRP as the enzyme catalyst for oxidation of luminol by hydrogen peroxide [108]. In this contribution, an integrated injector, separator, and postseparation reactor were fabricated on planar glass wafer. The fluorescein conjugate of HRP (HRP-F1) was used as a sample for optimization of the CL detector response. The schematic layout of the microchip is shown in Figure 16. Devices consisted of two pieces of 1.95-mm thick glass, one with etched channels and the other with drilled access holes, thermally bonded together. For some devices, aluminum mirrors were sput-

˚

ter-deposited to 1000 A thickness on the bottom plate after bonding. A shadow mask was formed with tape on the chip to define the 1-cm-x-1-cm-square mirrored region during deposition. When present, the mirror was centered on the Y- shaped reaction zone junction where the sample and peroxide stream met. Design PCRD1 was used for optimizing luminol and peroxide concentrations and evaluating the difference between doubleand single-T injection modes and design PCRD2 was used for optimizing PMT bias, PMT operating temperature, reaction pH, comparing lens numerical apertures, and evaluating the effect of channel depth. This design, with an integrated mirror, was used in immunoassay applications. The sample is placed in reservoir A, luminol in B, and H2O2 in D. Reser-