3-1 Nanowire chip selection in dry air condition
Nanowire Field Effect Transistor (NWFET) wafers were fabricated at Class 100
clean room of National Nano Device Laboratory (Hsinchu city, Taiwan, R. O. C.), sealing in an anti-electrostatic bag. We need to check the electrical properties before biosensing, by using electrical analysis machine (Keithley 2636) to measure NWFET chips. For the purpose, one standard operation process was built (Figure 9) to screen good
chips, record the data of Id-Vg plot and key parameters of threshold voltage, on/off
current ratio (Figure 10-A& 10-B), and then, look for good and stable chips for
biosensing. Chip selection principles were defined in Table 6, classified into three ranks,
best, acceptable and failed.
3-2 Biosensing of non-immobilized semiconductor device
Initially, the control set of conductance measurement was calculated by measuring the electrical variation of bare chip. The real-time changeable conductance was observing in all experimental preparation setup. First, the microfluidic channel was filled with 10 mM pH 7 phosphate buffers and kept the buffer flowing. The conductance value won’t change too much if we don’t do any action. When the syringe tube was changed, a big pulse wave appeared, caused by the little bubble
17
between two buffer solutions. These results show it didn’t change obviously because we just change the same solution to test whether the device is stable after the bubble pass through. Then, we injected 100 pM CA 16, the signal didn’t change obviously after the bubble peak passed because of no immobilization on nanowire device. After CA 16 DNA sample was injected, washed the microfluidic channel by pH 7 buffer before next sample. Then, we switched EV 71 DNA sample, the signal still be kept in baseline. Finally, we washed the channel by pH 7 buffer solution to clean the microfluidic channel, the signal still be kept in baseline at last. Those overall results show that the conductance electrical signal won’t change obviously without functionalizing nanowire device surface (Figure 11).
3-3 Device characteristic verification in liquid phase
After surface modification on NWFET chips, and the electrical characteristic of the device was tested. We want to know two main goals that the variation trend of the most sensitive voltage point is also based on the sensing theory. First, we swept the liquid gate voltage to check the electrical variation controlled by liquid and calculate these data. The conductance value (red line) (Figure 12) will be increased as liquid gate is increased and fitted the N-type device features. Then, differentiate the conductance value with liquid gate values to get Gm values. Finally, Gm values are
18
divided by conductance values (G) to get sensitivity values (Sensitivity=Gm / G , Gm = dG / dV). After those mathematic operations, we got the Vg value with maximum variation rate at sub-threshold region. Both biosensing and pH sensing shall be operated at fixed Vg value we calculated before. On the other side, we would check whether the variation trend of the conductance signals fitted its doping type. In this experiment, we applied N-type NWFET as sensing device, to observe the signal changeably fitting N-type characteristic. When the buffer switched from pH 7 to pH 8, it will have more negative ion than pH 7, to make N-type NWFET induced less electrons passing through those channels. Reversibly, buffer solution switched from pH 8 to pH 7, it has less negative ion than pH 8, to make N-type NWFET inducing more electrons pass through those channels. The conductance value is directly relative to the current of nanowire channels. As the results are shown in Figure 13, the conductance value could reversibly change from different pH buffers fitting N-type device characteristic.
3-4 Enterovirus 71 (EV 71) DNA Biosensing
After surface functional group modification and reaction with EV 71 DNA probes (Shown as Figure 6), we measured the real-time conductance signals of injecting buffer or samples in fixed flow rate by syringe pump (Kd Science). In the
19
experiment, the probe DNA was designed 20 base long and amino hexyl was modified at 5’-end with complementary DNA strand as target DNA but no modified at both ends.
In addition, we design a CA16 as a negative control to confirm the probe DNA of specifically binding to its’ complementary DNA. All samples were diluted by phosphate buffer (10 mM, pH 7.0) and we regard the buffer condition as baseline in the experiment. First, we would make a bubble to generate a signal pulse observing whether the baseline shift or not. So as to the other pulses when we switching to another samples, the bubble means to signal income. Second, it will be injected 10000 times concentration negative control (CA 16) than positive sample (EV 71) to observe whether the signal variation, then eluting the CA 16 with phosphate buffer. Third, inject the EV 71 1 fM into the channel to react with probe DNA. We can see those conductance value of 1 fM EV 71 obviously changed than 100 pM CA 16 as shown in Figure 14. Finally, we added 100 fM EV71 to check whether the reaction saturated, compared with those conductance changed percents among control and experiment data (Figure 15). As these results, the real-time conductance analysis method can specifically identify both complementary DNA and non-complementary DNA. The proposed measuring system has advantages like label-free, real-time and high sensitivity. It’s promising to develop semiconductor biosensor.
20
3-5 EV 71 DNA biosensing after hot water washed
After the EV 71 biosensing, we injected 10~15ml hot water (over 80 ℃) to remove the hybridized DNA sequence (Tm: 51.8 ℃). Then, washed the microfluidic channel by phosphate buffer (room temperature) and repeated previous process again.
First, we would sweep liquid gate to observe the conductance value still controlled by liquid gate because we didn’t know how the hot water affected the device characteristic. Secondly, got those data of conductance to liquid gate and calculate the most sensitivity Vg point. Compared to Figure 9, the most sensitive Vg point shifted to 1.65 V (Figure 16). We considered that (1) the device was decayed since the first biosensing process (2) after the hot water washed the device surface suffered damage.
However, we followed up previous process to do biosensing again. In the beginning, the baseline was rised up continuously, even we injected CA 16 (control set) DNA sample. But the conductance value will be decreased obviously when EV 71 (experiment set) injected. We got the same response even though the baseline was not as flat as these previous results. We can say that the conductance signal could truly reflected bio-reaction on device surface (Figure 17).
3-6 CA 16 DNA Biosensing
In contrast to EV 71 biosensing, we do another experiment by immobilizing
21
CA16 DNA probe and detecting complementary CA 16 target sequence. The EV 71 DNA strand became control set in the experimental design. According to the same previous process to sweep the Vg, the most sensitive Vg point was calculated as shown in Figure 18. We got the Vg value about 1.03 V and started biosensing by the fixed Vg value. In the beginning of CA 16 biosensing, we injected 100 pM EV 71 DNA sample as control set, and the conductance value still be kept on baseline. Then, injected 1fM CA 16 and the conductance value didn’t shift obviously. As injected 100 fM CA 16, the conductance value shifted down slowly. A little difference from EV 71 biosensing results, the conductance value would rise when we injected pH 7 PBS buffer to wash the channel. Similarly, we injected higher concentration of CA 16. It can obtain more variation than lower concentration samples, but still be recovered when buffer injected (Figure 19). After biosensing, we compared those different concentrations of CA 16 samples and control set (Figure 20), it was revealed the variation rate directly proportional to concentration. Finally, we proposed that the reason of the conductance value reversed when buffer injected. The distortion of microfluidic PDMS may change the streaming potential, to cause the hybridized DNA flushed by buffer. The higher concentration of CA 16 samples still be reacted with immobilized DNA probes. It still performed high specificity of biosensing because it didn’t react with control set.
22