3-1 Dual gate indium-gallium-zinc-oxide thin film transistor with an unisolated floating metal gate for threshold voltage modulation and mobility enhancement
In this study, we proposed a modified double gate a-IGZO TFT that can adjust the threshold voltage in both positive and negative directions without the additional dielectric layer and power supply. The control gate metal is formed directly on the back interface of the IGZO active layer. There is no dielectric layer between the IGZO body and the control gate, indicating a metal-semiconductor (MS) back gate. During device operation, the back gate is floated, and therefore the power supply for the back gate is not necessary. The back gate bias (VBG) is provided from the intrinsic built-in voltage across the IGZO body and the back gate. Because there is no dielectric layer beside the back gate, the control ability of the floating back gate is better than that of the conventional back gate formed by a metal-oxide semiconductor (MOS) diode. By choosing a floating back gate that processes work function higher or lower than that of IGZO (ψIGZO), we can significantly move the threshold voltage(from - 5.0 to 7.9 V).
3-1.1 Motivation
According to the experiment results mentioned before [29], threshold voltage influenced by light induced carriers at the back channel and about -7 volts shift was observed in transfer characteristic, as shown in Fig 3.1. That means threshold voltage is very sensitive to carrier concentration at the back channel. If capping materials with different work functions at the back channel, electrons can transfer inside or outside from the back interface due to the band
21
bending of a-IGZO to the capping material, threshold voltage could be modulated.
3-1.2 The threshold voltage varies with capping metallic layers with different work functions
The channel width and length are 1000 um and 300 um, respectively. Then, the 150-um-long metallic back gate was deposited on the IGZO top surface of some standard (STD) devices to form the double gate devices as shown in the inset of Fig. 3.3(a). The capping layer is located between the source and drain contacts. Various metal materials with different work functions include calcium (Ca, φ Ca=2.78 eV), titanium (Ti, φ Ti=4.33 eV), copper (Cu, φ Cu=4.65 eV), and gold (Au, φ Au=5.10 eV) were used as the floating back gate. Except Ca, the thickness of these capped layers is fixed as 60 nm. For Ca, to avoid oxidation, 100-nm-thick Al is passivated onto the 35-nm-thick Ca. The threshold voltage and mobility are extracted from the slope and the x-axis intercept of the √ID-VGS curve measured under saturation condition (VDS=20 V, VGS is scanned from -15 to 20 V). Fig. 3.3(a) presents the transfer characteristics of the uncapped (STD), Ti-capped, Ca-capped, and Au-capped a-IGZO TFTs.
As compared with the STD device, there are significant Vth shifts of -7.4 V and 5.5 V of the Ca-capped and Au-capped devices, respectively; a small Vth shift of -1.7 V of Ti-capped device is probed. Capping a metal layer (a control gate without insulator) on the IGZO back surface does not form a current leakage path to increase SS and leakage current. In Fig.3.3 (b), the slopes of √ID-VG curves are raised by metallic capping layers, indicating a significantly improved field effective mobility. Table. 3-1 lists the extracted typical parameters of a-IGZO TFTs with various metallic capping layers. A tunable Vth ranges from -5.0 to 7.9 V is demonstrated. Where the threshold voltage shift (ΔVth) is the threshold voltage difference between STD device and the floating dual gate (FDG) device. Besides, all these FDG devices
22
possess an improved field-effect mobility, a comparable SS (0.16–0.33 V/dec) and a high on/off current ratio (>2.6×108).
3-1.3 Mechanism of threshold voltage shift from metal capping.
The back gate bias (VBG) is contributed from the intrinsic built-in voltage between IGZO body and the floating back gate. As shown in Figs. 3.4(a) and (b), before contact, the Fermi-level in Au and Ca are lower and higher than that of IGZO, respectively. After IGZO contacts with Au, the thermal equilibrium is attained with a constant Fermi-level. The electrons in IGZO flow into Au to form a built-in voltage and a depleted IGZO body as shown in Fig.
3.4(c). The system can be regarded as a conventional dual gate TFT that has a control gate with a negative gate-to-source voltage (always off).The voltage is contributed from the work function difference,(ψIGZO−ψAu) e, where e is the electron charge. Because Au depletes the IGZO body, the channel formation on the dielectric will be suppressed and the Vth is increased (enhancement mode TFT). On the contrary, as shown in Fig. 2.4(d), Ca injects electrons into IGZO body. The injected electrons accumulate near the interface between IGZO and Ca to form a channel. The system can be regarded as a conventional dual gate TFT that has a control gate with a positive gate-tosource voltage _always on_. The voltage is contributed from the work function difference, (ψIGZO−ψCa)e. Because Ca generates a channel on the IGZO back surface, the Vth becomes more negative to suppress the initially existed channel
to turn-off the device (depletion mode TFT). Due to the thin active layer (e.g., 35 nm), the devices are operated with a fully-depleted IGZO body.[36],[37] The proposed double gate transistor can be regarded as a composition of three capacitors. They are the capacitor of the bottom main gate (CG, which is formed by the gate dielectric), the depletion capacitor (CD, which is formed by the depleted IGZO body) and the capacitor of the back gate (CBG, which is
23
formed by the MS back gate contact).[37] In the case of the enhancement mode operation, the channel forms on the dielectric surface and the body is depleted by the back gate. Therefore, the channel is located between CG and CD as shown in the bottom of Fig. 3.4(c). On the contrary, in the case of the depletion mode operation, the channel forms on the interface between IGZO and the back gate and the IGZO body is depleted by the bottom gate. Therefore, the channel is located between CD and CBG as shown in the bottom of Fig. 3.4(d). Under enhancement mode operations, the back-gate-voltage dependent threshold voltage shift (dVth) can be estimated by the formulas as: dVthCG=−dVBG[CDCBGCD+CBG]. Under depletion mode operation, dVth can be given as dVth[CGCDCG+CD]= −dVBGCBG.[37] Compared to the conventional dual gate TFT that uses a MOS diode as the back gate, the MS back gate contact in this study leads to a high capacitor (CBG). As a result, a small back gate voltage (VBG) can shift the Vth significantly. The derivation of Vth to back gate voltage, dVth dVBG, is as high as 5 for the proposed dual gate TFT.
3-1.4 Influence of a-IGZO thickness in dual gate structure
For the dual gate structure, the thickness of a-IGZO influences the control ability of main gate. In Fig 3.5(a), three standard devices with various a-IGZO thicknesses(15nm, 30nm, 50nm) capped with 80nm of Au, threshold voltage and mobility have strong dependence with a-IGZO thickness. As show in Fig 3.5(b), threshold voltage become more positive as a-IGZO thickness decrease indicates back gate becomes more dominant, and its cost higher voltage for main gate to turn on the device as shown in Fig 3.6 and parameters are listed in Table 3.2. On the contrary, as the a-IGZO thickness increase, it has barely threshold voltage shift after Au capped.
Mobility is also affected by the thickness of a-IGZO film, as thickness of a-IGZO decrease, higher mobility is measured. This trend might be explained as the reduction of scattering effect
24
in dual gate structure or a parasite back channel will be discussed in the next section.
3-1.5 Mobility enhancement and equivalent circuit of dual gate structure
In this study, 2 to 3 times increment of mobility has been observed, and the voltage between floating gate and source has been measured as shown in Fig. 3.7. As VGS larger then VTH, floating gate start to couple the voltage of main gate and induces the second channel under it. The equivalent circuit is shown in Fig. 3.8, and the voltage of floating gate, VDS×(R2/R1+R2),is influenced by its position. The limitation of mobility is due to two gaps of uncapped region. If the capping ratio increases, higher mobility value can be obtained.
3-1.6 An inverter comprised of an enhancement-mode and a depletion-mode a-IGZO TFT
An inverter comprised of an enhancement-mode a-IGZO TFT (Ti back gate) to serve as a switch and a depletion-mode a-IGZO TFT (Ca back gate) to serve as the load is demonstrated as shown in the inset of Fig. 3.9. The voltage transfer curve and the voltage gain of the inverter is shown in Fig. 3.9. With a supply voltage (VDD) of 20 V, a signal inversion behavior with a maximum voltage gain of -39 V/V is obtained. The maximum input voltage that will be recognized as a low input logic level (VIL) is 1.6 V. The minimum input voltage that will be recognized as a high input logic level (VIH) is 2.7 V. The output high voltage (VOH) is 20 V and the output low voltage (VOL) is 0.4 V. The transfer width, defined as VIH−VIL, is only 1.2 V. In this study, a dual gate IGZO TFT with a floating MS back contact is proposed to modulate the threshold voltage and to increase the field-effect mobility. The floating back gate has a
25
back-gate bias (VBG) contributed from the built-in voltage between the IGZO and the capping metal. By using various floating metals, a depletion mode or an enhancement mode dual gate TFT can be achieved. An improved mobility is also obtained in the proposed FDG TFT. An inverter comprised of the proposed dual gate a-IGZO TFTs has a maximum voltage gain of -39 V/V with a supply voltage of 20 V.
3-2 A novel approach to improve biochemical sensitivity of indium-gallium-zinc-oxide thin film transistor
(IGZO TFT) by capping sensing layer on active layer
Metal oxide semiconductor has many benefits beyond organic semiconductor for it stable in operation, low temperature fabrication, simple structure, less insensitive to oxygen and moisture than OTFTs, and can sensing in liquid surrounding. Base on those characteristics, metal oxide semiconductor is a possible candidate for next generation of biochemical sensor.
However, metal oxide semiconductor is less sensitive to biochemical particles than organic material, so a sensing layer can be added to solve this problem and sensitivity can be also improved. By alternating the sensing layer, different biochemical particles can be detected.
3-2.1 Electrical properties of IGZO capping different material
For detect certain gases, we chose P3HT( 3 - H e x y l t h i o p h e n e ) to detect ammonia and nitride oxide, and CuPC( C o p p e r p h t h a l o c ya n i n e ) to detect acetone.
To make sure that the capping materials does not cause the damage of electrical properties of a-IGZO film by measure transfer and output characteristics. Thickness influence of sensing material is also considered, and electrical characteristics of a-IGZO capped with different
26
thickness of P3HT and CuPC is measured. It shows that IGZO TFT keeps its field-effect control, and no obvious decrement of mobility has been observed. For P3HT capped device, threshold voltage varies with P3HT thickness , shown in Fig. 3.10 , can be attribute to the heating process to solidify P3HT solution in glove box. Glove box is full of N2, therefore oxygen is tend to escape from the surface from IGZO film under heating process, and it is reasonable that thicker P3HT capped a-IGZO TFT has lesser threshold voltage shift. For CuPC capping, deposited under thermal evaporation, does not exhibit large threshold voltage shift and mobility variation after capping as shown in Fig. 3.11. All parameters are shown in Table 3.3.
3-2.2 Ammonia sensing properties of P3HT capped IGZO TFT
The responses of standard (STD) a-IGZO TFT and P3HT-capped a-IGZO TFT to ammonia are investigated. Drain current variations (I) of these two devices are plotted as a function of time exposed to different ammonia concentration in Fig.3.12. Devices are biased at VGS-VTH=5V and VDS=20 V. For STD device, no significant response can be observed when ammonia concentration is 10 ppm. For P3HT-capped device, an obvious current drop is obtained when device is exposed to ammonia. After the removal of ammonia, the current drop is recovered. The ammonia sensing response of a P3HT-based OTFT has been reported with ammonia concentration ranging from 10 to 100 ppm. [38] In that report, ammonia molecules behave as acceptor-like deep trap states (or as electron donors) to trap the holes at the P3HT/dielectric interface, shift the threshold voltage to be more negative and cause a current drop. In our study, a positive threshold voltage shift is observed when device is exposed to ammonia as shown in the Fig. 3.14. In our work, a clear response to 0.2 ppm ammonia is observed as shown in Fig 3.15. The sensing sensitivity is also strongly influenced by the
27
thickness of P3HT. When the thickness of P3HT increases to be 70 nm, the response to 10-ppm ammonia becomes almost invisible as shown in the in Fig 3.13.
3-2.3 Nitride oxide sensing properties of P3HTand PbPC capped IGZO TFT
A typical oxidizing gas, nitric oxide (NO), is used to stimulate the P3HT-capped device.
Drain current variations (I) of STD device and of P3HT-capped device are plotted as a function of time exposed to different NO concentration in Fig.3.16. Devices are biased at VGS-VTH=8.6 V and VDS=20 V. When standard device is exposed to NO gas, a slow current drop is observed. The current drop is not able to be recovered even when NO is removed. The slow and irreversible response is due to the slow adsorption and desorption of NO molecules on the oxide semiconductor thin film. [32]. When P3HT-capped a-IGZO TFT is exposed to NO, on the contrary, a fast current increase is observed. A fast recovery behavior is also obtained when NO is removed. The P3HT capping successfully blocks the reaction between NO and a-IGZO film. The sensing behavior is dominated by the reaction between NO and P3HT in which NO molecules act as oxidizing agents to withdraw electrons from P3HT.[39] The positive potential on P3HT film helps to turn on the transistor and thus increase the current of a-IGZO TFT.
3-2.4 Acetone sensing properties of CuPC capped IGZO TFT
Copper phthalocyanine (CuPc) is capped onto the back channel of a-IGZO TFT to detect a kind of reducing gas, acetone, which is an index gas in the breath of patients with metabolic disease like diabetes mellitus.[31] Drain current variations (I) of standard device and CuPC-capped device are plotted as a function of time exposed to different acetone
28
concentration in Fig. 3.17. Devices are biased at VGS-VTH=0.5 V and VDS=1 V. For standard device, no clear response to acetone can be observed, indicating a weak charge transfer interaction between the a-IGZO and acetone at room temperature. For CuPC capped a-IGZO TFT, on the contrary, a fast, sensitive, and reversible response to acetone is obtained. The response current variation ratio (IDS/I0 when sensing time is fixed as 120 sec) is plotted as a function of acetone concentration in Fig. 3.18 A linear relationship is observed. The sensing sensitivity is as high as 100 ppb and is promising to be used to distinguish healthy humans (
900 ppb) and diabetes patients ( 1800 ppb).
3-2.5 Sensing mechanism of hybrid IGZO gas sensors
The organic layer and the a-IGZO film form a p-n junction. Oxidizing or reducing vapor molecules act like electron acceptors or electron donors to change the potential of the organic sensing layer and thus change the energy band equilibrium of the p-n junction. As a result, the current of a-IGZO TFT is significantly changed. The proposed mechanism has been utilized to form a sensitive visible light sensor in previous report. [29] In that work, visible light absorbed by the organic capping layer produces electron-hole pair, injects electrons into a-IGZO film, and changes the TFT threshold voltage. In another work, we capped various kinds of metals onto the back interface of a-IGZO TFT. The built-in potential between the floating capping metal and a-IGZO is utilized to adjust the device threshold voltage. The capping layer is served as a floating second gate to influence the characteristics of a-IGZO TFT as shown in the Fig.
3.19. When the organic sensing layer (OSL) is exposed to oxidizing or reducing gases, the potential of the OSL is changed. Since the OSL is treated as a floating second gate, the potential variation of the OSL significantly influences the current of the a-IGZO TFT.
Ammonia molecules absorbed onto P3HT act like reducing agents to inject electrons into
29
P3HT. The negative potential on P3HT layer, the second gate, shifts the threshold voltage to be more positive and produces a current drop in a-IGZO TFT. The sensing sensitivity is high because the threshold voltage is sensitive to the body potential.
3-2.6 The relationship between sensitivity and gate bias
An interesting gate-bias-dependent sensing sensitivity is also found when changing the bias conditions during sensing. As shown in Fig. 3.20, the current variation ratio (IDS/I0) is plotted as a function of time when devices are exposed to 5-ppm acetone and recovered in pure nitrogen under different bias condition. When VGS-VTH=5 V and VDS=1 V, IDS/I0 is less than 0.04 and is similar to IDS/I0 of the floating gate condition, implying that the sensing mechanism is similar to a chemical resistor rather than a chemical transistor. When VGS-VTH=0.5 V and VDS=1 V, IDS/I0 is improved 5.5 times and reaches 0.22. The gate-bias-dependent sensitivity can be explained as follows. When VGS is large, the channel carriers are mostly induced and controlled by VGS. The potential variation of the OSL (i.e. the second gate) caused by the charge transfer between acetone and the OSL is much smaller than VGS. Thus, the channel current has only a weak response to the acetone molecules. When VGS-VTH=0.5 V, the potential variation of the OSL is comparable to VGS-VTH. The channel current is then significantly influenced by the charge transfer between the OSL and gas molecules.
30
3-3 High mobility a-IGZO TFT with Ca/Al capping layer
In the previous experiment, we capped different metals on the back side of a-IGZO TFT, shown in Fig. 3.3, and an interesting result has been found. The mobility of Ca/Al capped devices is always higher than others, and more the 10 times mobility enhancement can be reached by Ca/Al capping.
In this section, a set of experiments have been demonstrated to investigate the high mobility effect (about than 10 times increment) found in Ca/Al capped a-IGZO TFT. The comparison of Ti capped and high mobility Ca/Al capped devices are shown in Fig.3.21, and listed in Table 3.4.
3-3.1 Transfer characteristics and time decay of Ca/Al capped TFT
After Ca/Al capping, we trace the device for 50 days, as shown in Fig 3.22. Variation of threshold voltage and mobility during 50 days is shown in Fig 3.23. As Ca/Al capped, threshold voltage became negative. To explain the threshold voltage shift, the work function of Ca is 2.8 eV, much higher than the Fermi level of a- IGZO (~ 4.5 eV). And the shift of threshold voltage can be explained by the electrons injection from calcium into a-IGZO due to the work function difference as a back gate discussed in section 3-1. A negative gate bias is needed to deplete the active layer and to turn off the device. When devices are exposed to air, the rapid oxidation of Ca eliminates the threshold voltage shift. An explanation includes two mechanisms is proposed for mobility surge compare to cap with other metals (eg. Al, Ti, Au) demonstrated in section 3-1. The first mechanism is that, for metal oxide semiconductors, oxygen vacancies can be regarded as doping forms highly conductive regions cause high mobility, and variation of oxygen concentration in a-IGZO film due increment of oxygen vacancies at the edge and under Ca capping layer due to the formation of Ca-O bonds.
31
Gradually decrement of mobility after exposed to air can be explained as the reduction of oxygen vacancies in oxygen rich surrounding.
The second mechanism is defect reduction in a-IGZO film after eliminated weak-bonded oxygen by oxidation of Ca, by forming Ca-O, and that may reduce the shallow traps in metal oxide semiconductor and lower energy barriers which limit electrons transport. After 30 days, the device has threshold voltage around -0.5 and mobility about 90cm2/VS, the device
becomes stable and keeps high mobility. Detail parameters are listed in Table 3.5.
3-3.2 .Stability test of Ca/Al capped IGZO TFT
Stability is also an important issue to know if Ca/Al capping causes degradation of a-IGZO film, which limits this work in practical application. A positive bias stress (PBS) and negative bias stress (NBS) is demonstrated. (VG-VT=20 for PBS and VG-VT=-20 for NBS, and VD was not supplied during bias stress.) The transfer characteristics and threshold voltage shift as shown in Fig. 3.24. After same bias time, similar voltage shift for Ca/Al capped and standard
Stability is also an important issue to know if Ca/Al capping causes degradation of a-IGZO film, which limits this work in practical application. A positive bias stress (PBS) and negative bias stress (NBS) is demonstrated. (VG-VT=20 for PBS and VG-VT=-20 for NBS, and VD was not supplied during bias stress.) The transfer characteristics and threshold voltage shift as shown in Fig. 3.24. After same bias time, similar voltage shift for Ca/Al capped and standard