Chapter 1 Introduction
1.4 Organization of this Thesis
In Chapter 2, we present the fabrication of the n-type α-IGZO TFTs by describing the basic process flow The measurement setups employed for device characterization are also presented in this chapter.
In Chapter 3, we present and discuss the electrical characteristics of the fabricated devices, including the on-off current ratio, µFE, SS and Vth. The impacts of the deposition power and oxygen flow on the device characteristics will be discussed respectively. Also, we investigate the effects of various post treatments on the
electrical characteristics of the devices. Effects of an n+ layer inserted between the α-IGZO channel region and source/drain metal contact are also studied.
Finally, we summarize the concluding remarks from our experimental results and the suggested future work in Chapter 4.
10
Chapter 2
Device Fabrication and Measurement
Setup
2.1 Manufacturing Process of IGZO Targets and Characteristics of Deposited IGZO Films
In this study, thecomposition of the IGZO powder used in the experiment was mixed by an atomic ratio of In:Ga:Zn = 1:1:1
.
The target powder (In2O3, Ga2O3, ZnO) with high purity was purchased from Optotech Materials Co. The powders were then milled, followed by calcining at 1000°C for 1 hour. Next, the powders were ground and sieved through a 250-mesh screen. Afterwards, the material was further formed into the sputter target under a pressure of 60,000-70,000 kg/cm2 at 1250°C for 60 minutes in Ar ambient.After the IGZO target was formed, we mounted it in the sputter chamber for the deposition of IGZO thin films on glass substrates. Table 2.1 shows the major process parameters of the deposition. Certainly the process conditions may affect the properties of the deposited material, such as the atom ratio of deposited α-IGZO film and the deposition rate. In this work, we used n&k analyzer to measure the thickness of IGZO thin film. The atom ratio of the deposited α-IGZO films was investigated by X-ray photoelectron spectroscope (XPS). The surface roughness was investigated by atom force microscope (AFM). From Fig 2.1, we can see that higher deposition power has faster deposition rate owing to the higher sputter yield. However, the mean free
path of the sputtered species could be reduced as a higher oxygen partial pressure is implemented and, as a result, the deposition rate is slower. Form the X-Ray Diffraction (XRD) material analysis of IGZO films which was deposited in RT, we found out that the IGZO film without post annealing is always in amorphous state regardless of the deposition conditions.
2.2 Device Fabrication and Process Flow
In the section, we introduce the basic process flow of the α-IGZO TFT devices.
The device structure belongs to the inverted-staggered type, which is the most commonly used structure for active matrix liquid crystal display (AMLCD). Figure 2.2 shows the schematic process flow for fabricating the devices. It can be seen that the α-IGZO TFT device was fabricated on the surface of a silicon dioxide film thermally grown on a silicon substrate. For fabricating this structure, a 100 nm Al-Si-Cu was first deposited by physical vapor deposition (PVD) as gate electrode and patterned by wet etching (Fig. 2.2(a)). Then, a 100 nm TEOS oxide was deposited by plasma-enhanced chemical vapor deposition (PECVD) as the gate dielectric (Fig.
2.2(b)). It is worth noting that before depositing the α-IGZO active layer, we had to pre-sputter the target for 15 minutes to prevent the contamination of IGZO target surface. Subsequently, a 50 nm α-IGZO film was deposited as the channel layer by the RF sputter at room temperature under different deposition power and oxygen partial pressure (Fig. 2.2(c)). The atomic ratio of IGZO target was In:Ga:Zn=1:1:1.
The system’s base and working pressures were 3×10-6 Torr and 5 m Torr, respectively.
During sputtering, the flow rate of Ar was fixed at 50 standard cubic centimeter per minute (sccm), while the oxygen flow was set at 1sccm, 3sccm, or 5sccm, with oxygen partial pressure of 9.8 ×10-5, 2.83 ×10-4 or 4.54 ×10-4 Torr, respectively, all
12
with a fixed deposition power (100W). The oxygen partial pressure was derived from O2/(Ar+O2) × working pressure. In this work, we also explored the effect of the RF power by setting the power at 100, 125, 150, 200W at constant Ar/O2 gas mixing ratio (50/1) and oxygen partial pressure of 9.8 ×10-5 Torr. After the α-IGZO channel film was deposited, an optional n+α-IGZO layer and a source/drain metal were deposited using lift-off process (Fig. 2.2(d)). The optional n+ α-IGZO layer allows us to compare the electrical characteristics of devices with and without it. Such an n+ α-IGZO film was deposited without flowing the oxygen during sputtering. The source/drain metal was a 300 nm Al/Ti metal compound which was deposited with deposition power fixed at 75W by a DC sputter and the working pressure was 3 m Torr (Fig. 2.2(e)). Afterwards, we performed a lithographic step to define the active device region (Fig. 2.2(f)). Since the etching rate of the α-IGZO with hydrochloric acid (HCl) was too fast, we used a diluted HCl solution (HCl: H2O = 1:200) instead to avoid damage and severe lateral etching of the α-IGZO channel film (Fig. 2.2 (g), (f)).
In order to contact the gate electrode, contact etching was performed by wet etching using buffer oxide etcher (B.O.E.). Fig. 2.2 (i) and Fig. 2.2 (j) show the cross-sectional views of the α-IGZO devices without and with the n+ α-IGZO layer.
2.3 Measurement Setup
In our study, electrical measurements of all devices were executed by an HP4156A precision semiconductor parameter analyzer, and the measurement temperature was maintained at 25oC. Prior to the measurements, all α-IGZO TFTs samples used in this study were annealed at 200 oC for 40 minutes on the hot plate aiming to remove the excess moisture on TFTs.
The basic electrical parameters of the fabricated device were extracted from the
electrical characteristics. Note that the threshold voltage (Vth) was not extrapolated from the Id-Vg curves, rather, the Vth was defined in this work as the value of Vg when Id equals 1nA ×
under Vd of 0.1V, where W and L are the channel width and the channel length, respectively, as estimated from the patterns of the mask. The subthreshold swing (SS) was calculated by the following equation:
SS ! $ "#
% , (Eq. 2-1) the minimum SS value was extracted in the drain current region between 10-9A to 10-12A. The transconductance (Gm) was also extracted from the Id-Vg curves, and we could calculate the linear µFE by the following equation with VD=0.1V:
µ'( ·*+
·,-.·$/ . (Eq. 2-2) For on-current (Ion), the value was defined as the ID when VG – Vth = 8V at VD=10V.
14
Chapter 3
Results and Discussion
3.1 Effects of Deposition Conditions
3.1.1 Introduction
α-IGZO channel layers are usually deposited by sputters because of the low deposition temperature and thus the feasibility for manufacturing of devices on flexible substrates. It is well known that the properties of amorphous metal oxide semiconductor materials are strongly dependent on the processing conditions. In this regard, there had been several reports exploring the effects of deposition power [29], oxygen flow [18, 29] and working pressure [18], on the basic electrical characteristics of α-IGZO TFTs. The analysis of different process parameters are the major object of this section with an intent to acquiring suitable process parameters for deposition of α-IGZO layers with superior properties. In particular, the impact of different sputtering parameters is investigated.
3.1.2 Effects of Deposition Power
Figures 3.1(a)~(d) show the transfer characteristics of the α-IGZO devices deposited with deposition power of 100W, 125W, 150W and 200W, respectively, at a constant oxygen flow of 1sccm. In each figure at least 10 devices were measured in order to understand the uniformity of the characteristics. Overall the uniformity of the manufacture process is good as can be seen in the figures. Typical device characteristics of each split are put together in Fig. 3.2 for comparison. Some clear
trends are shown in the figures. First, the transfer curve is shifted to the left with increasing sputtering power. Second, a higher deposition power results in larger off-state current. The extracted Vth, SS, and on-current for each split are shown as a function of channel length in Figs. 3.3(a), (b) and (c), respectively. In Fig. 3.3(a) it is seen that the variation of Vth is reduced with increasing deposition power, though the trend of SS seems not clear in Fig. 3.3(b). As shown in Fig. 3.3(c), the on-current measured at the same gate overdrive is higher as the deposition power increases. It is known that the In3+ cations in α-IGZO channel film would extend the conduction band minimum and reduce the effective mass of the conduction electrons which would enhance µFE and reduce the resistivity of the α-IGZO film [17].
Moreover, the Vth is negatively shifted by increasing the carrier concentration.
However, the carrier concentration of α-IGZO channel film is controlled by oxygen vacancies and the incorporated In concentration.
Table 3.1 shows that the value of In/(In+Ga+Zn) ratio of the deposited films which increases with increasing deposition power. The concentration of these metallic compositions was obtained with the XPS technique. The dependence of the In/(In+Ga+Zn) ratio on the deposition power is postulated to have a strong influence on device characteristics including Vth, µFE, carrier concentration and off current [17, 42]. However, as shown in Fig. 3.3(b), the SS is slightly deteriorated at a higher deposition power. This indicates that the interface traps between gate oxide and α-IGZO channel layer are not seriously affected by the deposition power. Fig. 3.4 shows that the field-effect mobility (µFE) extracted from the transconductance enhances with increasing deposition power. The µFE for the devices processed at a lower deposition power is slightly reduced as the channel length is shorter than 20μm.
It is postulated to be related to the effect of the S/D parasitic resistance in these
16
devices with a shorter channel length. Next, this issue is clarified by measuring the resistance components in the devices.
The total resistance (Rtotal) as a function of channel length can be evaluated by using the total resistance method conducted in the linear region [24]:
R1 12 $!/
/ µ
345·,-.· $6$78# R9:, (Eq. 3-1) where Cox is the oxide capacitance, W is the channel width, L is the channel length, µE
is the effective mobility and RSD is the S/D parasitic resistance. From this equation, we can evaluate µE and channel resistance per micrometer as a function of gate overdrive, and the results for devices with L/W = 100/400μm characterized at VD= 0.1V are shown in Fig. 3.5(a). The extracted resistance components are shown in Fig.
3.5(b). The measured RSD values are in the range from 3.5 to 6.7 kΩ, more than one order in magnitude smaller than the channel resistance in this case. On the other hand, the results for a device with a short channel length of 10 μm are shown in Fig. 3.6. In this case the extracted RSD values are comparable to that shown in Fig. 3.5(b).
However, due to the ten times smaller in channel length, the channel length is significantly reduced. As a result, the impact of the RSD on device characteristics is no longer negligible in the short-channel devices.
In Figs. 3.7(a) and (b), it is seen that the channel resistance decreases while µE
increases, respectively, with increasing gate overdrive voltage. It is obvious that a higher gate overdrive voltage induces more free electrons to transport in α-IGZO channel layer which is the main reason for the reduced channel resistance. On the other side, the more induced electrons would reduce the potential barrier for carrier conduction [32, 34-37]. As has been introduced in Chap. 1, the electrical behavior of the α-IGZO TFTs can be explained by a percolation model. As shown in Fig. 1.4(b), as the Fermi level is manipulated by the gate bias in the degenerate state, shorter
carrier transport paths around the valley of energy barrier could be formed as the gate voltages increases. The output characteristics of the devices with different deposition power are shown in Fig. 3.8. The differences among the splits in on-current at the same gate overdrive voltage are obvious.
3.1.3 Effects of Oxygen Flow
Figures 3.9(a) and (b) show the transfers characteristics of several α-IGZO devices with oxygen flow of 3sccm and 5sccm, respectively, at a constant deposition power of 100W. As compared with the results shown in Fig. 3.1(a) for the split with oxygen flow of 1sccm, the variation in device characteristics is not worse. Fig. 3.10 shows the typical transfer curves of the three splits. As described in Chap. 1, oxygen vacancies are the main source to release free electrons to transport in the metal-oxide semiconductors. This is clearly confirmed in the figure that the transfer curves are positively shifted with increasing oxygen flow. Therefore, a higher gate voltage is necessary to accumulate free electrons to form a conductive layer between the source and drain. In addition, the extracted electrical parameters are show in Figs.
3.11(a)~(d). Fig. 3.11(a) shows that the Vth has a weak dependence on the channel length. This is because the channel length is more than 30x longer than the gate oxide thickness, thus the short channel effects are expected to be small. In Fig. 3.11(b), the SS increases with increasing oxygen flow. It is known that the SS can be affected by the bulk defects in the α-IGZO channel film in addition to the interface traps, thus the reduction in SS might result from the greater densification and lower amount of the bulk defects as deposited at a lower oxygen partial pressure [18, 29, 43]. Fig. 3.11(c) shows the on-current measured under the same gate overdrive voltage of 8V. From this plot, we know that on-current reduces with increasing oxygen flow owing to
18
deficiency in carrier concentration. Fig. 3.11(d) shows that µFE is dependent on the channel length. The electrical behavior is similar to that described in pervious sub-section and RSD should have drawn effect on the short-channel devices.
The channel resistance and µE extracted from total resistance method are shown as a function of gate overdrive in Fig. 3.12. It is obvious that a higher gate overdrive voltage induces more free electrons to transport in the channel, which is the main reason to reduce channel resistance. Because the deposited α-IGZO channel layer with a higher oxygen flow has less carrier concentration, a higher potential barrier height and longer transport path are expected from the percolation model (Fig. 1.14 (a)). Therefore, the channel resistance and µE are higher in the films with a lower oxygen flow. Fig. 3.13 shows the output characteristics of devices deposited with different oxygen flow. Despite the higher resistivity, the ohmic S/D contacts are still formed for the split with high oxygen flow as shown in the output characteristics, attributed to the large S/D contact area. However, the undesirable RSD effects still need to be taken into consideration and solved in practical circuit applications in which the contacts are minimized for performance concern. In order to solve this problem, several methods have been proposed, such as plasma treatment [25-27], post-annealing treatment [28-29] and post-laser annealing [30]. In the next section, we explore an approach which adopts an n+ insertion layer to improve the device characteristics.
Summary of the major effects of the deposition parameters on the electrical characteristics of α-IGZO TFTs are shown in the Table 3.2. Based on these observed trends, most of the devices characterized in the remaining sections of this thesis were fabricated with the following suitable sputtering parameters: deposition power is 150W, Ar/O2 gas mixing ratio is 50/1, oxygen flow is 1sccm. Major electrical
characteristics of the fabricated devices are as follows: the Vth is 0.66±0.15V, SS is 0.476±0.023V/decade, µFE extracted by transconductance is 10.22±0.36 cm2/Vs, and the on-current is 2.09±0.03×10-5A at VG-Vth=8V and VD=10V.
3.2 Effects of the Insertion n
+Layer between Source/Drain and the Channel
Deposition conditions of the n+ insertion layer are described in Chap. 2. It is a heavily-doped α-IGZO layer which has a high carrier concentration in order to reduce the contact resistance of the S/D metal contacts. To confirm this point, we took the Hall measurements to probe the carrier concentration and resistivity of the deposited films. The resistivity of the n+ α-IGZO layer is 2.85×10-2 Ω-cm and the carrier
concentration is 1.28×1019cm-3. Details about the device fabrication are also given in Chap. 2. To reduce interface defects between the semiconductor layer and metal, the n+ layer and the S/D metal were deposited sequentially in a sputter with base pressure of 2 ×10-6 Torr.
Fig. 3.14 shows the output characteristics of the α-IGZO devices with and without n+ layer. From the output performance, we can see that the on-current of α-IGZO devices with n+ layer is increased by 36% at the same gate overdrive. The transfer characteristics of the two devices are shown and compared in Fig. 3.15. As compared with the device without the n+ insertion layer, both on and off currents increase with the n+ insertion layer. The increase in on-current is attributed to the reduction in the parasitic resistance and is consistent with the results presented previously [44]. The main reason is attributed to the thinning of the tunneling width at the S/D contacts due to a high amount of carrier concentration [33]. However, the
20
causes for negative shift of the threshold voltage and increase in the off current are not clear at this stage, presumably due to the difference in film thickness because of the lack of reproducibility with the homemade sputtering system. Figures 3.16(a), (b), and (c) show µFE, Vth, and on-current of the two splits of devices. Effects of the insertion layer in improving the mobility and on current are clearly illustrated in the figures.
3.3 Effects of Different Annealing Ambient
The ambient for the post-annealing treatment of the fabricated devices was varied and studied in order to understand the effects on the characteristics of α-IGZO devices. The post annealing process is one of the major factors to dictate the electrical behavior because this process can generate or diminish free carriers in α-IGZO channel films and modulate the film’s properties. In this study, the annealing ambient employed includes the forming gas (95%N2/5%H2) and nitrogen environment, and the annealing duration is one hour. Fig. 3.17 shows the representative transfer curves after the annealing treatment and Fig. 3.18 compares the extracted electrical parameters.
The on-current measured at VG-Vth=8V at VD=10V are 2.01±0.03×10-5A, 1.78±0.09×
10-5A and 1.66±0.03×10-5A for fresh devices and annealed ones done in forming gas and nitrogen gas, respectively. As can be seen in Fig. 3.17 that the transfer characteristics of α-IGZO TFTs are dramatically affected by the annealing ambient. In order to explain these results, we postulate some possible schemes and explore their suitability for the observed electrical behaviors of the annealed samples.
In the α-IGZO films the metal species tend to bond with the oxygen ones. As oxygen vacancies are generated inside the films, more electrons are released. When the α-IGZO structure is annealed in a higher temperature, it generates oxygen vacancies and it is well known that one oxygen vacancy can release two free
electrons.
From the on-current performance, we find that the on-current of devices receiving post annealing treatment is smaller than that of fresh ones. It is postulated that the carrier concentration is reduced by the post annealing process and obviously the film’s properties are dramatically affected by the treatments. Accounting for this point, three different possible conditions for the channel films are shown in Figs.
3.19(a) ~(c). Figure 3.19(a) corresponds to the situation of fresh devices in which a homogeneous film is assumed. In Figs. 3.19(b) and (c), we assume that, during the post-annealing treatment, the hydrogen and/or nitrogen atoms presenting in the environment may diffuse into the α-IGZO channel film from the back channel surface and modify the film with a high resistivity. However, in Fig. 3.19(b) it is assumed that thickness of such a modified resistive region is thinner than the original film thickness, while in Fig. 3.19(c) the whole film is modified. From Fig. 3.20, we find that the off-state current is not dependent on gate voltage but on drain voltage. Such a phenomenon implies that the performance of the off-current is related to intrinsic
3.19(a) ~(c). Figure 3.19(a) corresponds to the situation of fresh devices in which a homogeneous film is assumed. In Figs. 3.19(b) and (c), we assume that, during the post-annealing treatment, the hydrogen and/or nitrogen atoms presenting in the environment may diffuse into the α-IGZO channel film from the back channel surface and modify the film with a high resistivity. However, in Fig. 3.19(b) it is assumed that thickness of such a modified resistive region is thinner than the original film thickness, while in Fig. 3.19(c) the whole film is modified. From Fig. 3.20, we find that the off-state current is not dependent on gate voltage but on drain voltage. Such a phenomenon implies that the performance of the off-current is related to intrinsic