Chapter 2 Theory Description
2.3 The drift phenomenon of ISFET
2.3.2 Expression for drift
According to Ref. [16], we can know the two kinds of the drift mechanisms, the surface effect and the bulk effect to the various sensing layers. The concept of the trap of hydrogen-bearing species can be seen as the application of the site-binding model.
In 1998, Jamasb proposed a physical model quantitatively to explain drift in terms of the bulk effect, hydration of the silicon nitride surface [17].
It is known that the drift phenomenon of an ISFET is modeled by implementing a hopping transport mechanism, known as dispersive transport [18] [19], to determine the rate of hydration of the sensing layer. We can see the dispersive transport in a broad class of disordered materials. Dispersive transport leads to a characteristic power-law time decay of diffusivity [20] which can be described as:
1
00 0
( ) ( )
D t =D ωt β− (2-22) where D00 is a temperature-dependent diffusion coefficient which obeys an Arrhenius relationship,ω0 is the hopping attempt frequency, and β is the dispersion parameter satisfying 0<β < . In turn, dispersive transport also leads to decay in 1 density of sites/traps occupied by the species undergoing transport [20]. This decay is described by the stretched-exponential time dependence given by:
/ ( ) / (0) exp[( / )]
It is important to keep the biasing current constant while operating an ISFET. We
can achieve the purpose easily in the normal condition. But the sensing layer is getting hydrated all the time when the electrode is immersed in the aqueous electrolyte. The hydration makes the chemical composition of the sensing layer different from that of the bulk of the sensing material. Consequently, the dielectric constant is dissimilar to that of the un-hydrated layer. The equivalent insulator capacitance is also determined by the series combination of the surface hydration layer and the underlying sensing film, will exhibit a slow, temporal change. When the ISFET is biased in the region we need, the gate voltage will simultaneously exhibit a change to keep a constant drain current. The change of the gate voltage can be written as [21]: where the Q is the inversion charge. If the pH, temperature, and the ionic strength inv of the solution are kept constants, Eref , Φ , Si Ψ , and0 χsol can be neglected, so the drift an be rewritten as:
1 1
In our study, the gate oxide of the fabricated ISFET was consist of two layers, a lower layer of thermally-grown SiO2 of thickness, tL, and an upper layer of
sputter-grown ZrO of thickness, 2 t . U CI(0) is the effective insulator capacitance given by the series combination of the thermally-grown SiO2 capacitance, εL tL, and the sputter-grown ZrO capacitance, 2 εU tU . C t is analogous to i( ) CI(0), but an additional hydrated of capacitance make Ci always smaller than CI, εHL tHL, at the oxide-electrolyte interface must be considered, and the sputter-grown ZrO 2 capacitance is now given by εU [tU −tHL] . The series combinations of the capacitance are illustrated in Fig. 2-6. Therefore, the drift is given by:
( ) U HL ( ) (2-29) can be seen as constant value no matter what the substrate is. According to this assume, it is possible to eliminate the drift or hold the drift to be a constant at any
According to Eq. (2-32), we can know that if the immersing time of the gate oxide is long enough, the shift of gate voltage will approach almost a constant value which is greatly dependent on the hydration depth, tHL( )∞ .
2.4 References
[1] Y. Q. Miao, J. R. Chen and K. M. Fang, “New technology for the detection of pH”, J. Biochem. Biophys. Methods, vol. 63, pp. 1-9, 2005.
[2] P. Bergveld, “ISFET, Theory and Practice”, in IEEE Sensor Conference, Toronto, Oct. 2003.
[3] H. Kaden, H. Jahn, M. Berthold, ”Study of the glass/polypyrrole interface in an all-solid-state pH sensors”, Solid State Ionics, vol. 169, pp. 129-133, 2004.
[4] P. Bergveld, “Thirty years of ISFETOLOGY What happened in the past 30 years and what happen in the next 30 years”, Sensors and Actuators B, vol. 88, pp. 1-20, 2003.
[5] H.K. Liao, et al. ”Study on pHpzc and surface potential of tin oxide gate ISFET”, Materials Chemistry and Physics, vol. 59, pp. 6-11, 1999
[6] R.E.G. van Hal et al. , “A general model to describe the electrostatic potential at electrolyte oxide interface”, Advance in Colloid and Interface Science, vol.69, pp.
31-62, 1996.
[7] Miao Yuqing , Guan Jianguo, Chen Jianrong, “Ion sensitive field transducer-based biosensors”, Biotechnology Advances, vol. 21, pp. 527-534, 2003.
[8] D. E. Yates, S. Levine, and T. W. Healy, “Site-binding model of the electrical double layer at oxide/water interface”, J. chem. Soc. Faraday Trans. I, vol. 70, pp.
1807-1818, 1974.
[9] W. M. Siu, R. S. C. Cobbold, “Basic Properties of the Electrolyte-SiO2-Si System:
Physical and Theoretical Aspects”, IEEE Transactions on Electron Device, vol.
ED-26, NO. 11, Nov., 1979
[10] 吳浩青, 李永舫, ”電化學動力學”, 科技圖書公司, 2001 年 2 月 [11] Tadayuki Matsuo, Masayoshi Esashi, ”Methods of ISFET Fabrication”, Sensors
and Actuators, vol. 1, pp. 77-96, 1981.
[12] Imants R. Lauks, Jay N. Zemel, “The Si3N4/Si Ion-Sensitive Semiconductor Electrode ”, IEEE Transactions on Electron Devices, vol. ED-26, no.12, pp.
1959- 1964, Dec., 1979.
[13] J. C. Chou, C. Y. Weng, “Sensitivity and hysteresis effect in AL2O3 gate pH- ISFET”, Materials Chemistry and Physics, vol. 71, pp. 120-124, 2001.
[14] P.D. van der Wal et al. ,”High-K Dielectrics for Use as ISFET Gate Oxide”, in Sensors, Proceedings of IEEE.2004
[15] H. K. Liao et al., ”Study of amorphous tin oxide thin films for ISFET applications”, Sensors and Actuators B, vol.50, pp. 104-109, 1998.
[16] Luc Bousse, Piet Bergveld, “The Role Of Buried OH Sites In The Response Mechanism Of Inorganic-Gate pH-Sensitive ISFETs”, Sensors and Actuators, vol. 6, pp. 65-78, 1984.
[17] S. Jamasb, S. D. Collins, and R. L. Smith, ”A Physical Model for Threshold Voltage Instability in Si3N4-Gate H+-Sensitive FET’S (pH ISFET’s)”, IEEE Transactions on Electron Devices, vol. 45, no. 6, pp. 1239-1245, Jun, 1998.
[18] G. Pfister, H. Scher, ”Time-dependent electrical transport in amorphous solid:
As2Se3”, Physical Review B, vol. 15, no. 4, pp. 2062-2082, Feb., 1977.
[19] H. Scher, Elliott W. Montroll, ”Anomalous transit-time dispersion in amorphous solid”, Physical Review B, vol. 12, no.6, pp. 2455-2477, Sep., 1975.
[20] J. Kakalios, R. A. Street, W. B. Jackson, ”Stretched-Exponential Relaxation
Arising from Dispersive Diffusion of Hydrogen in Amorphous Sillicon”, Physical Review Letters, vol. 59, no.9, pp. 1037-1040, Aug. 1987.
[21] S. Jamasb, S. D. Collins, R. L. Smith, ”A Physically-based Model for Drift in Al2O3-gate pH ISFETs ” in International Conference on Solid-State Sensors and Actuators Chicago, June, 1997.
[22] S. Jamasb, S. D. Collins, R. L. Smith, ”A physical model for drift in pH ISFET ”, Sensors and Actuators B, vol. 49, pp. 146-155, 1998.
Chapter 3
Experiment and Measurement
3.1 Introduction
The structure of ISFET is similar to the conventional MOSET. The most difference between MOSFET and ISFET is the gate electrode. The ISFET takes the gate membrane as a sensing layer immersed in the pH-solution [1]. In the experiment, we produce the n-type ISFET and the p-type ISFET. From the derivations of the ISFET sensitivity in chapter 2, there are no factors about electrons and holes. So we can guess that the sensitivity must be the same whether the substrate is n-type or p-type.
Besides the ISFET, we also produce the REFET. REFET is identical with ISFET except that REFET does not react on the ion concentration to be measured, i.e.
REFET is insensitive to the test ion concentration [2]. We use the polymer materials coated on the ZrO2 sensing layer to get the insensitive REFET. And then, we measure the electric parameters including the drift amount and the sensitivity of ISFET and REFET respectively.
3.2 The fabrication process flow of ISFET and REFET
The all procedures of experiment are accomplished in NDL (National Nano Device Laboratory) and NFC (Nano Facility Center). The fabrication process flow is shown in Fig. 3-1. And the detailed steps are listed in this section.
Part 1:
1. RCA clean.
2. Wet oxide growth 6000Ǻ, 1050°C, 65mins.
Part 2:
3. Mask#1 for the definition of the regions of Source/Drain.
4. Wet etching of unblocked silicon dioxide.
5. Dry (Screening) oxide growth 300 Ǻ, 1050°C, 12mins.
6. Source/Drain implantation:
Dose= 5*1015 (1/cm2), Energy= 15Kev (Boron) for p-type ISFET.
Dose= 5*1015 (1/cm2), Energy= 25Kev (Phosphorus) for n-type ISFET.
The Above detailed information is in Table 3-1.
7. Source/Drain annealing, 950°C, 60mins.
Part 3:
8. PECVD Oxide deposition 1µm.
Part 4:
9. Mask#2 for the definition of the contact hole and gate region.
10. Wet etching of unblocked silicon dioxide.
11. Dry oxide growth 100 Ǻ, 850°C, 60mins.
Part 5:
12. Mask#3 for the definition of the sensing layer region.
13. Sensing layer (ZrO2) deposition by Sputtering 300 Ǻ.
14. ZrO2 sintering 600°C, 30mins.
Part 6:
15. Mak#4 to define the contact region.
16. Al deposition by sputtering 5000 Ǻ.
Part 7:
17. Backside Al evaporation 5000 Ǻ.
18. Al sintering 400°C, 30mins Part 8:
19. Coating the polymer-based material as the REFET sensing layer.
3.3 The illustration of the core parts
In part 1, RCA clean is for the elimination of ions and clean for the silicon surface, to make the growing equality of Wet Oxide better.
In part 2, the area of the defined Source/Drain grow a layer of 100A screen oxide to protect the surface when ion-implantation. The other use of this layer of oxide is avoid the channeling effect while implantation. After ion-implantation, the doped ions with high energy destroy the structure of lattice sites of the silicon substrate, and repair it by the way of anneal.
In part 3, then using PECVD grow oxide 1 mµ as insulation to prevent buffer solution effect the device when ISFET is immersed in the electrolyte.
In part 4, decide the gate oxide region, and then use BOE etching the unblocked oxide. It grows the better quality gate oxide100A with dry oxide.
In part 5, define the sensing layer region for growing 300A ZrO2 cause its property of the high sensitivity, which is about 57mv/pH [3]. Table 3-2 is the sputtering parameters.
In part 6, deposition of 5000A Al for the defined electrode region.
In part 7, finished Backside Al evaporation 5000A is the primary process of ISFET.
In part 8, coating the polymer materials cause of its insensitivity onto the ISFET sensing layer.
Before step 19, we coat HMDS onto the ZrO2 sensing layer to enhance the
adhesion of the interface between the ZrO2 sensing layer and the insensitive polymer material.
In step 19, an important step in the process of REFET fabrication, we coat the polymer-based material onto the HMDS layer as the REFET sensing element. We use a three-layer structure to form the REFET sensing region. The HMDS and the specific polymer-based material are coated from bottom to top onto the ZrO2 sensing layer.
These materials conclude epoxy, the mix composition of phtoresistance (PR) and Nafion (NF), and the different ratio of the mixture of Nafion and polyimide (PI). The experiment test structures are listed in Table 3-3.
According to the data, Nafion (shown as Fig.3-2) is extremely resistant to chemical attack and able to be used in relatively high temperature, so Nafion can protect for the underlying layers from damaging. There is also an additional advantage for using Nafion. It is that Nafion is a cation exchange polymer, i.e.
NFafion will not affect the hydrogen ions to pass and any sensitivity decreased. The coating procedure has to be controlled carefully, following is our process flow:
1. Prepare Epoxy, the mixture of NF and PR (FH6400) with the ratio of NF/PR
= 1/1, the mixture of NF and PI with the ratio of NF/PI = 1/1 and 1/3, and the pure PI.
2. Dropping the above materials onto the HMDS above the ZrO2 sensing layers. Dried in air for 30 hours, these materials become colloid with thickness of 5-10um.
3.4 Measurement system
3.4.1 Preparation before measurement
We obtain the I-V curves of the pH-ISFET from the measurement using HP4156 as measurement tool and the system (shown in Fig 3-3) to investigate the characteristics of the regular ISFETs and the REFET structures with the different polymer-based sensing layers. To reduce the disturbance from the environment, such as light influence, the entire measurement procedures were executed in a dark box [4].
The measured pH values are 1, 3, 5, 7, 9, 11, 13, and the pH buffer solution were supplied by the Riedel-deHaen corp.
3.4.2 The measurement of pH sensitivity
A HP-4156 semiconductor parameter analyzer system were set up to measure the I-V characteristic curves, in which include IDS-VGS and IDS-VDS curves at some specific controlled temperature. We must be careful while dropping the pH-buffer solution at the sensing region. Because of the small sensing areas, it could easily generate the air bubbles in the interface. Consequently, in order to measure in the more stable situation, we began to measure after about 30 seconds while the sensing layer getting immersed in the pH buffer solutions.
From IDS-VGS curves, we can extract the pH sensitivity (in the unit of mV/pH) of ZrO2 pH-ISFET. First, we should decide the specific bias IDS usually at the point of the maximum transconductance, i.e. the maximum slope of IDS-VGS curve. The sensitivity of pH-ISFET, the gate voltage difference between different pH value buffer solutions at the constraint of the chosen current IDS in the IDS-VGS curve, is obtained.
The illustration to find the selectivity is the sketched in Fig. 3-4.
3.4.3 The measurement of drift
To obtain the more accurate drift amount, we put the n-type and the p-type ISFETs respectively immersed in the buffer solution with pH 3, 5, 7, 9, 11 in the periods of 12 hours before measuring. Similarly, we put the REFETs into the pH 7 buffer solution in the period of 90 minutes before measuring. The REFET breaks easily so that we take the less immersion time. As the similar manner in measuring the selectivity, we can sketch the gate voltage versus time in 400 minutes with the equal intervals of 10 minutes. The illustration to find the drift is the sketched in Fig. 3-5.
3.5 References
[1] P. Bergvled,”Development of an ion-sensitive solid-state device for neurophysiological measurement”, IEEE Trans. On Bio-Med. Eng. (1970) 70-71
[2] P. Bergveld, “Thirty years of ISFETOLOGY What happened in the past 30 years and what happen in the next 30 years”, Sensors and Actuators B, vol. 88, pp. 1-20, 2003.
[3] K. M. Chang, K. Y. Chao, T. W. Chou, and C. T. Chang, ”Characteristics of Zirconium Oxide Gate Ion-sensitive Field-Effect Transistors” Japanese Journal of Applied Physics Vol. 46 No. 7A pp. 4334-4338 2007.
[4] Paik-Kyun Shin, ”The pH-sensing and light-induced drift properties of titanium dioxide thin films deposited by MOCVD”, Applied Surface Science, vol. 214, pp.
214-221, 2003
Chapter 4
Results and Discussions
4.1 Introduction
The drift is an unavoidable phenomenon of pH-ISFETs, some studies already indicate that the gate voltage shift according to the immersed time goes on. It makes the inaccuracy of the output signal and the lower reliability. We will discuss the experiment results and propose a method to reduce the influence. In addition to drift, we also list the selectivity of p-type and n-type ISFET.
In section 4-2, the important properties of REFET such as the drift and the sensitivity are investigated. From the properties, we can choose the suitable inactive material as the REFET sensing layer.
4.2 The p-type and n-type ISFET
We choose ZrO2 as a sensing film of pH-ISFET for the reason of the higher sensitivity. Fig. 4-1 and Fig. 4-2 individually represent the measured IDS-VGS curves from pH 1 to pH 13 of the p-type and the n-type pH-ISFETs. The sensitivity is 57.8 mV/pH for the p-type ISFET and 58.73 mV/ pH for the n-type ISFET. The bias conditions of the two ISFETs are all arranged in Table 4-1. The high sensitive ZrO2
gate ISFET in the readout circuits could accomplish the high revolution while determining the pH value. Besides of the higher sensitivity, the ZrO2 film also includes other advantages [1].
Additionally, we focus on the drift phenomenon of the p-type and the n-type
ZrO2 pH-ISFET. According to Ref. 2, we know the drift phenomenon is a complex function with the variables such as containing time, sensing film quality and pH aqueous solution. In order to obtain the more precise experiment results, the time and the film must be controlled at same condition in the different pH aqueous solutions.
Fig. 4-3 ~ 4-7 are the gate voltage with times for 400 minutes drift in the intervals of equal 10 minutes to the pH 3, 5, 7, 9, 11 aqueous solution of p-type ISFET. Similarly, Fig. 4-8 ~ 4-12 show times for 400 hours drift in the pH 3, 5, 7, 9, 11 aqueous solution of n-type ISFET. From these results, we can find that the drift is a time dependent function which is obeyed with the study investigated by S. Jamasb [3]. The model presented by hydration S. Jamasb indicates that the hydration is a continuous process. The immersion in the solutions only diminishes the drift amount. The action can not vanish the phenomenon.
We also find that the drift phenomenon is related with the pH aqueous solution from Fig. 4-13 and Fig. 4-14. The drift comparison between p-type ISFET and n-type ISFET are listed in Fig. 4-15 and Fig. 4-16. The drift rate increasing with pH value is obeyed with prior research [4]. From our measurements, the direct proportion between the drift rate and the pH value almost exists no matter the categories of ISFET. The possible mechanism of the result is discussed in the following. OH- plays a major role in the hydration action. In the alkali solutions, the more hydroxyl ions make the hydration more. So the drift rates increase with pH value.
From the experiment results, a possible method is proposed to reduce the drift influence at the region of pH 5 – pH 11. In the n-type ISFET, we define the real gate voltage N = n - ∆n. n represents the gate voltage of the ideal n-type ISFET, and -∆n is the drift amount of the n-type ISFET. We can know that n and ∆n are both positive numbers. So the signal to noise ratio is indicated as S/N|n-type ISFET = n/∆n. In the p-type ISFET, we can similarly find the S/N|p-type ISFET = p/∆p. p and ∆p are both
positive numbers. To reduce the drift effect, we can define a new variable T = N – P.
From the above discussions, we can find T = (n+p) – (∆n-∆p). And then the S/N ratio is calculated as S/N|T = n + p/∆n - ∆p. By comparing the three S/N ratios, we know that S/N|T is larger than S/N|n-type ISFET and S/N|p-type ISFET. The comparison between the original S/N and the modified S/N is listed in Table 4-2. From Table.4-2, the proposed method can enhance the S/N ratio. In other words, we can reduce the drift influence with the introduction of the new variable T.
4.3 The results in the REFET structure
We coat the HMDS film to improve the adhesion between the ZrO2 layer and the selected materials. Because of the focus is other materials in this experiment, we do not hope the additional function provided from HMDS. Fig. 4-17 is the selectivity of the structure of HMDS/ZrO2. It shows that HMDS is not the major factor in the reduction of the sensitivity. The transconductance comparison between the ZrO2 layer and the HMDS/ZrO2 structure is shown in Fig. 4-18. The transconductances are almost the same. From Fig. 4-17 and Fig. 4-18, HMDS is an acceptable choice for improving the adhesion.
The parameter, transconductance (gm), is an important property for the decision whether a material was chosen as the REFET sensing layer. It is troublesome to design two different circuits to maintain the reference bias point if the transconductance of the specific material is not similar with that of the ZrO2 gate ISFET. The second reason for the same transconductance is the essentiality to select the REFET material. REFET is identical with ISFET except the low pH sensitivity.
Usually, we take ISFET and REFET to combine together with a differential OP amplifier. The output signal is only associated with the pH value after the elimination
of the common mode signal of the differential OP amplifier. If the transconductance is extremely different, the output voltage is related with many factors. Therefore, the transconductance of materials is an essential factor which can help us to make an appropriate decision.
At the beginning of this experiment, we investigated epoxy, the mix composition of photoresistant (PR) and Nafion (NF) and the mixture of polyimide (PI) and Nafion.
Fig. 4-19 shows the comparison of their transconductances. Fig. 4-19 implies that epoxy is not suitable in the situation. We can only use the two mixtures. Because of
Fig. 4-19 shows the comparison of their transconductances. Fig. 4-19 implies that epoxy is not suitable in the situation. We can only use the two mixtures. Because of