In our previous work (data not shown), current responses to hydrogen peroxide of the needle electrodes fabricated with different conducting materials, such as a pure graphite powder sealed electrode, a mixture of MWCNTs and graphite powder sealed electrode, a pure MWCNTs sealed electrode, were studied. It was found that the needle-type electrode showed the highest response to H₂O₂ when the ratio of graphite and carbon nanotubes up to 1:2 (w/w). This optimized mixture of MWCNTs/graphite powder, together with different percentages of G_ox powder, were packed into the glass capillary to construct the glucose needle-type biosensors. The current responses of this biosensor to the presence of different glucose concentrations were then studied (see Figure 1). As indicated from this figure, the current response of biosensor increased with the percentage of G_ox. At certain range of G_ox, a larger amount of G_ox in the biosensor correspond to a higher current response. When the composition ratio of MWCNTs mixture/G_ox is increased from 5:3 to 5:5, the intensity of current response was not increased so obviously. This indicates that when the percentage of G_ox is increased to a certain level, the corresponding increase of current response to the percentage of G_ox become small. The reason is that the decrease of percentage of MWCNTs results in an increase of the amount of G_ox in the whole electrical material, therefore the electrical conductivity and hydrogen peroxide catalysis of the biosensor will be decreased. This decrease can counteract the effect of G_ox in the biosensor, thus the biosensor can't display a sustained increase in current response along with the increase of G_ox amount.

In order to study the stability and reproductivity of the needle-type biosensors, three biosensors with different ratios of MWCNTs mixture and G_ox were tested in the presence of 20 mM glucose solution for 14 days (Figure 2). The biosensors were removed from a 4°C refrigerator and put in the reaction cell to perform the experiment each day, and then dried and stored at 4 °C again after each experiment was completed. During the continuous observation, the MWCNTS mixture/G_ox biosensor with a composition ratio of 5:3 showed the best reproductivity and stability. After 14 days observation, three biosensors all showed a decay trend; the current response of the biosensor with a ratio of 5:3 had the smallest decay and the current response of the biosensor with a ratio of 5:1 decreased deeply (the current response of the 14th day was only 65 % to the result of 1st day). Due to the continuous observation, the biosensors had to pass through a dry to wet to dry again cycle. This cyclic procedure had a negative influence to the activity of enzyme, and normally the activity of enzyme will decrease and hence the current response of biosensor; but in our investigation, these decays were not so obvious over a long period. This can be explained by the integration of enzyme protein with MWCNTs: the acid treated MWCNTs have a lot of hydrophilic carboxyl groups, which are easily combined with enzyme protein (G_ox) in the moist environment and form stable enzyme-MWCNTs complexes [13]. These complexes are more stable and result in a long-term stability of the G_ox compared with the free G_ox. Therefore this kind of biosensor showed a better current response and stability.

To test whether or not this needle-type biosensor based on carbon nanotubes can be used in a Continuous Glucose Monitoring system, we put the biosensor with a composition ratio of 5:3 (MWCNTs mixture/G_ox) in a micro reaction cell containing glucose solution and observed its current responses continually.

First, the needle biosensor was equilibrated in the PBS (pH 7.0) for 3600 s (1-hour). Then the current response of the biosensor was measured in PBS for another 3600 s (1-hour) (at the applied potential of +0.6V vs. SCE). Later, we pumped glucose solution into the micro reaction cell continually to maintain a constant glucose concentration of 5 mM, and then recorded the current response for 23 hours. The total observation time is 24 hours (86400 s) and the results are shown in Figure 3. As shown in this figure, when the glucose concentration changed from 0 to 5 mM, the current response of the needle biosensor rapidly jumped from 18 nA to 45 nA. It was found the baseline of the biosensor decreased very slowly during the monitoring period, so after eliminating this baseline influence, the decay of the current response was less than 10% during the 24 hour online monitoring. This indicates that the needle biosensor has acceptable long-term stability. This result makes it possible to employ such kinds of CNT biosensor in the continuous glucose monitoring system field. In addition, the baseline problem could be solved by using two identical microbiosensors together, one is used as a glucose sensor and the other is used as a baseline reference sensor. Software changes in future use should result in another more efficient method.

A attractive amperometric biosensor for Continuous Glucose Monitoring has been presented, based on the MWCNTs/graphite/Gox packed needle-type electrode. The glucose biosensor sensitivity was strongly influenced by the glucose oxidase concentration within the MWCNTs mixture. After optimization, the biosensor displayed good sensitivity and stability, and a detected range of up to 20 mM. We also assessed the use of needle-type biosensor in a real-time glucose monitoring system. Our preliminary results indicate that the new needle-type biosensor is promising and useful in this field (current decay is lower than 10% during 24 hour observation). In addition, because all the experiments were processed in vitro, the system could not provide all the significative results we are interested in, so animal experiments will be considered for future work.

A CHI800 Electrochemical Workstation (American CH Instrument Company) was used., Glucose oxidase (G_ox, 100U/mg) was obtained from Sangon Bio Inc. All other reagents were of the best quality available commercially. Deionized water was used throughout. Glass capillaries (inner diameter is 0.5mm) were purchased from American Drummond Scientific Company. Multi-Wall Carbon Nanotubes (MWCNTs) were provided by the College of Materials Sciences and Chemical Engineering, Zhejiang University. The MWCNTs had an average length of 20 μm and a mean diameter of 15 nm. The MWCNTs were treated with mixed acid (H₂SO₄-HNO₃ = 3:1, v/v), and ultrasonically agitated for 8 hours. The acid treated MWCNTs were filtered and then washed with deionized water until neutral, and finally dried in an oven.