【正文】 but cannot be directly adopted in the chipAFS system. In the chipMS system, the makeup solution can be steered to the expected direction under highvoltage’s control, while in the chipAFS system, the makeup solution wasdelivered with the peristaltic pump and a backpressure at the junction with the separation channel would be produced. For this reason,another type of makeup solution channel, one quarter of a circle ( diameter) was directly etched on the chip and the makeup solution channel was tangent to the separation channel, so that the flow of the makeup solution was in the same direction as that of the EOF in the separation channel and the hydrodynamic effect of the makeup solution flow on the EOF was decreased. The channel dimension between junctions 6 and 7 was purposely enlarged to have significantly lower resistance to flow than the separation channel and also to reduce thenegative effect of makeup solutionon EOF.. Consideration of the chipCE–AFS interfaceThe efficiency of the chipCE separation should not be impaired by the interface. Thesample must be transported to the detector as efficiently as possible in a form that can beread by the AFS. As can be seen in Fig. 1, a concentric “tubeintube” design was employed to couple the chipbased separation system with the AFS [31]. A 100μL of Eppendorf pipette (intermediate tube) was cut into an appropriate length (25 mm), into which a piece of Teflon tubing (. mm, inner tube) was inserted. One end of the Teflon tubing was inserted into the chipexit port 7, which was manually enlarged to acmodate theTeflon tubing,and epoxysealed on the side wall of the chip. The big end of the pipette was plugged with an appropriate silicone gasket, into which the other end of the Teflon tubing was inserted until the end of the Teflon tubing reached a position approximately – to the small end of the small end of the pipette was inserted into a thicker wall silicone tube (outer tube) tightly. Two holes were opened at appropriate places on the walls of the intermediate tube and the outer tube to introduce the KBH4 solution and argon flow into the interface, respectively. The chip effluent (the mixture of separation channel effluent and the makeup solution HCl)merged with KBH4 solution at the exit of the inner tube. The reaction mixture was swept by an argon flow into the GLS,where the volatile arsenic species were separated from the F. Li et al. / J. Chromatogr. A 1081 (2005) 232–237 235 mixture and detected by the AFS. A 20% (v/v) solution of HCl and % (m/v) KBH4 were supplied separately by the two peristaltic pumps of a model FIA3100 flow injection analyser (Vital Instrumental Co. Ltd., Beijing, China). Obviously,the KBH4 solution and argon flow forced the chip effluent into the GLS, so the effect of the backpressure due to hydrodynamic effect on the separation was minimized.The position of the outlet of the inner tube was found to be critical for the chipCE–AFS interface. When the outlet of the inner tube was placed beyond the intermediate tube, poor peak shape and repeatability were observed. The optimum position of the outlet of the inner tube located just – inside of the intermediate tube. In this place, the effluent from the inner tube can react with the KBH4 efficiently and the generated hydrogen is easy to release forward.The hydrostatic pressure resulting from the elevation difference between the liquid surface in the reservoirs and the chip exit would also exert an effect on the liquid surface in the reservoirs was higher (～3 mm) than that inthe chip exit, so that the sample had a potential to flow to the chip exit even when there was no electrical field. This hydrostatic pressure would shorten the analysis time, but might impair the separation efficiency. To offset the effect of the hydrostatic pressure, the elevation of the GLS was adjusted to where the liquid surface in the sidearm of the GLS was level with that in the reservoirs. Compared with other chipbased system [28,29], the reservoirs in this work were not made airtight so that the liquid surfacewas easy to adjust,and this was vital for good repeatability.. Factors affecting chipCE separationThe migrations of ions in chipCE generally occur underthe bined action of electrophoretic and electroosmoticflow. Applying positive high voltage on the buffer reservoir 3,EOFis oriented toward the makeup solution reservoir 4, while the electrophoretic mobility of the anions, such as As(III) and As(V), is in opposite direction. If the magnitude of the EOF is higher than electrophoretic mobilities, arsenic species will consequently move towards the makeup solution reservoir 4. This is usually the case for As(III). However, As(V) was not detected at all even under high EOF conditions because of its high electrophoretic mobility. Thus, it is necessary to suppress or reverse the direction of EOF.It is well known that some surfactants could reverse the EOF [35,36]. So, a cationic surfactant, CTAB was added into the electrolyte buffer to modify the channels surface,which led to the appearance of the positive fixed charges and reversed EOF. Separation of As(III) and As(V) was achieved by reversing the polarity of the electric field in this system.The effect of the concentration of CTAB on the separation was examined in the concentration range from to mmol l?1. It was shown that no apparent difference wasobserved on the separation with the variety of CTAB this work, mmol l?1 CTAB was included in the buffer solution.The pH of the electrolyte buffer is one of the keys controlling chipCE separation. It influences the separation characteristic by affecting the electrophoretic mobility of the arsenic anions as well as the EOF. A pH range of – for the buffer solution (25 mmol l?1 H3BO3 and mmol l?1 CTAB) was investigated. It was found that As(III) and As(V) could be baseline separated in a pH range of –, while best separation efficiency was achieved at a p