In recent years, many reports interests in the formation of SiOx film at low temperature (≦ 200 °C) for fabricating electron devices on flexible plastic substrates.
[11-13] Among various deposition methods, the physical vapor deposition (PVD) is favorable as a result of the advantages of simple process, low cost, and conformity with low-temperature fabrication. Nevertheless, a high temperature annealing or plasma treatment is generally taken as post-treatment to passivate the traps in PVD-deposited film for improving the dielectric characteristics. Due to the limit of glass transition temperatures (Tg), these high temperature post-treatments are unsuitable to plastic substrates,6 and a low temperature treatment to passivate traps is required.
the SCCO2 fluid is proposed to fabricate the resistance memory at 150 °C, so it would be applicable to product RRAM on the substrates with low glass transition temperatures (Tg). The supercritical fluid technology is generally used for the impurity extraction, dehydration and drying of materials with a fine structure. It also had employed SCCO2 fluid to deliver oxidant into metal oxide film for terminating electric defects by the gas-like and liquid-like properties. According to the mechanisms of bistable resistance states are possibly dominated by some kind of charge trap, the we would employ SCCO2 technology to fabricate a resistance memory by varying the trap density in SiOx film. The experimental works will focus on the effect of passivating traps by SCCO2 treatment and investigate the influence of varying trap density on resistance switching phenomena.
Chapter 2
Application of Supercritical Fluid Technology on Silicon-Oxide Dielectric Thin Film
2.1 Fabrication of
Silicon-Oxide
and Experiment ProcessThe e-gun system was used to deposit SiOx films on p-type (100) silicon wafer under 2 ×10-6 torr at room temperature, and the average thickness of SiOx films measured by ellipsometer system is 5-7 nm. For enhancing the quality of these ultra-thin SiOx films, three different post-treatments were applied individually. First method, labeled as “H2O vapor treatment”, is immersing the SiOx film into a pure H2O vapor ambience at 150 °C for 2 hrs, in a stainless steel chamber. Second method, labeled as “3000 psi-SCCO2 treatment”, is placing the SiOx film in supercritical fluid system at 150 °C for 2 hrs, where was full of 3000 psi-SCCO2 fluids mixed with 8 vol.% ethyl alcohol and 2 vol.% pure H2O. The ethyl alcohol acts surfactant between nonpolar SCCO2 fluids and polar H2O molecules to making the uniform distribution of H2O molecules in SCCO2 fluids, and the SCCO2 fluids thereby could transport H2O molecules efficiently into SiOx layer to react with traps. In comparison with 3000 psi-SCCO2 treatment, the third method is treating SiOx film with 3000 psi pure SCCO2
fluids but no co-solvent was added, and labeled as “SCCO2-only”. The supercritical fluid system is shown in Fig. 2-1.Afterward, the Al metal was thermally evaporated on the top surface of these treated SiOx films and the backside of silicon wafer as electrodes to shape the metal insulator semiconductor (MIS) structure. The Fourier transformation infrared spectroscopy (FTIR) was applied to determine the evolution of chemical functional bonding after different treatment. The atomic content of SiOx film was detected from Auger electron spectroscopy (AES) analysis, and the electrical behaviors of SiOx film were measured by HP 4156-A semiconductor analyzer and
Agilent 4284A CV meter. The experiment processes of thin SiOx film with various treatments are exhibited in Fig. 2-2.
2.2 Analysis of Material and Discussion
2.2.1 Fourier Trans-form Infrared Spectroscopy (FTIR) Analysis
The FTIR spectra of SiOx films after different treatments are shown in Fig. 2-3, and the un-treated SiOx film is taken as background. For SCCO2-only treated SiOx film, there is no IR absorption peak appears in the wavenumber of 400-1500 cm−1, i.e. the SCCO2 fluid wouldn’t affect the functional structure of SiOx. In the case of H2O-vapor treated SiOx film, however, vibration bands for Si–O stretching at 1080 cm−1, Si=O stretching at 1230 cm−1, Si–O rocking at 440 cm−1, and S-H bending at 610 cm-1 increase clearly. [14]The raise of IR absorption peak indicates that the H2O molecules could infiltrate into SiOx film and modify the quality of SiOx film during H2O-vapor process. Because the main variation occurs in S-O bonding, it is believed that the H2O molecule is operative oxidant to react with Si dangling bond, and this agrees with prior literatures. [15~16] The same IR absorption peaks are also observed for 3000-psi SCCO2 treated SiOx film, and the higher absorption peaks express that SCCO2 fluid added with co-solvent (8 vol.% ethyl alcohol and 2 vol.% H2O) is a more operative method than H2O-vapor to enhance the quality of SiOx films. Therefore, the SCCO2
fluids mixed with co-solvent would completely infiltrate in into SiOx films as the pressure increasing in 3000psi, and the best improvement for SiOx films is achieved because of advancing the reaction probability between H2O molecule and Si dangling bonds. , and the transporting mechanism for SCCO2 fluids taking H2O molecule into SiOx film is shown in Fig. 2-4
2.2.2 Thermal Desorption System – Atmospheric Pressure Ionization Mass Spectrometer (TDS-APIMS) Analysis
The TDS measurement, as shown in the Fig. 2-5, was carried out upon heating these treated SiOx films from 50 to 800 °C at a heating rate of 10 °C/min in vacuum (10−5 Pa.). In Fig. 2-5 (a), m/e (mass-to-charge ratio) = 18 peak that is attributed to H2Owas monitored to evaluate the content of residual moisture in SiOx films. It is clearly found the same H2O content is detected in the SCCO2-col, H2O-vapor and 3000 psi-SCCO2-treated SiOx film, certainly consistent with the FTIR observation.
From Fig. 2-5 (b), m/e (mass-to- charge ratio) = 44 peak that is attributed to CO2, the residual CO2 in SiOx is equal the same after 3000 psi-SCCO2 , H2O-vapor and pure SCCO2 treatment. This is result from SCCO2-col, , H2O-vapor and pure SCCO2
treatment mean that no residual moisture and CO2 in the SiOx film. Therefore different Electrical Characteristics were not induced by residual moisture and CO2.
2.2.3 X-ray Photoelectron Spectroscopy (XPS) Analysis
XPS involves measuring the photoelectron spectra obtained when a sample surface is irradiated with x-rays. The kinetic energy (peak position) of the photoelectrons can be written as
-E - -q h
EK B s
where h is the x-ray energy, EB is the binding energy (the difference between the Fermi level and the energy level being measured), is the work function of the s electron spectrometer, q is the electronic charge, and is the surface potential.
We have also performed XPS measurements using an Al Kα X-ray source (1486.6 eV photons) to determine the bonding environments of the Si and O atoms.
the XPS spectra for Si 2p level that were calibrated from C 1s peak at 284.5 eV. Each spectrum was represented the result at different post-treatments, including SCCO2-only,
H2O vapor and 3000psi-SCCO2 treatment. The first group labeled as pure SCCO2
treatment, was designed as the control sample, and was only is treating SiOx film with 3000 psi pure SCCO2 fluids but no co-solvent was added,s. The second group labeled as H2O vapor treatment, was immersed into a pure H2O vapor ambience at 150 °C for 2 hrs in a pressure-proof stainless steel chamber with a volume of 30cm3. The third group marked as 3000psi-SCCO2 treatment, was placed in the supercritical fluid system at 150°C for 2 hrs, where was injected with 3000psi of SCCO2 fluids mixed with 5 vol.% of ethyl alcohol and 5 vol.% of pure H2O. As shown in Figure 2-6 , the Si2p peak, which have binding energies of 103.9 and 99 eV, respectively related to Si-O bonding and Si=Si in SiOx. However, the binding energy of Si2p peak shown in Figure 2-6 varied from 103.8 eV for pure SCCO2 fluids sample to 104.1 eV for 3000psi-SCCO2 treatment sample. The origins of binding energy shift (ΔBE) are suggested as a number of factors such as charge transfer effect, presence of electric field, environmental charge density, and hybridization. Among these, charge transfer is regarded as a dominant mechanism causing a binding energy shift. According to the charge transfer mechanism, removing an electron from the valence orbital generates the increment in core electron’s potential and finally leads a chemical binding energy shift [17]. Therefore, it is considered that the Si2Ppeaks shift originated from the enhanced charge transfer with different post-treatments, i.e., the larger portion of Si atoms was fully oxidized with 3000psi-SCCO2 treatment. Figure 2-7 shows the O 1s core level peaks also demonstrated binding energy shift with changing of different post-treatments. Each peak can be split into two sub-peaks by Gaussian fitting which represent the O-Si bonding at ~533 eV [18,19]. The peak intensity of O-Si bond for different treatments is not the same, meaning that these post-treatments would make different influence on the thickness and quality of the interfacial SiOx film. For the H20 vapor SiOx film, however, the peak intensity of Si-O bands raises apparently in comparison with the pure SCCO2 fluids -treated SiOx film. This is believed well that the H2O vapor would permeate into SiOx film and makes reaction with Si dangling bonds (i.e. traps) forming Si-O bands. [20~22]. These traps in the low-temperature
deposited SiOx film could be thereby passivated by H2O vapor molecules. Furthermore, with SCCO2 treatment, obvious increase in the intensity of Si-O bonding is observed in the XPS. It indicates that the best transport efficiency of H2O molecules into SiOx film is achieved by the SCCO2 fluids, potentially modifying the dielectric properties of SiOx
film, and the transporting mechanism for SCCO2 fluids taking H2O molecule into SiOx
film is shown in Fig. 2-4. Summary of binding energies for SiOx films are shown in Table 2-1 and Table 2-2.
2.2.4 Auger Electron Spectroscopy (AES) Analysis
In order to analyze the composition of the silicon oxide film after various post-treatments, including pure SCCO2 fluids, H2O vapor and 3000 psi-SCCO2
treatment, we performed the Auger electron spectroscopy analysis. The first group labeled as pure SCCO2 fluids treatment, was designed as the control sample, and was only treated in pure SCCO2 fluid at 150 °C for 2 hrs. The second group labeled as H2O vapor treatment, was immersed into a pure H2O vapor ambience at 150 °C for 2 hrs in a pressure-proof stainless steel chamber with a volume of 30cm3. The third group marked as 3000psi-SCCO2 treatment, was placed in the supercritical fluid system at 150°C for 2 hrs, where was injected with 3000psi of SCCO2 fluids mixed with 5 vol.%
of propyl alcohol and 5 vol.% of pure H2O. The ethyl alcohol plays a role of surfactant between nopolar-SCCO2 fluids and polar-H2O molecules, so that the H2O molecule uniformly distributes in SCCO2 fluids and be delivered into the SiOx film for passivating defects. As shown in Figure 2-8, the pure SCCO2 fluids-treated and H2O vapor-treated films has oxygen composition lower than that of the silicon oxide film after 3000 psi-SCCO2 treatment. The ethyl alcohol plays a role of surfactant between nopolar-SCCO2 fluids and polar-H2O molecules, so that the H2O molecule uniformly distributes in SCCO2 fluids and be delivered into the SiOx film for passivating defects.
2.3 Analysis of Electrical Characteristics and Discussion
2.3.1 The current density-electric field (J-E) characteristics
The plot of leakage current density of SiOx films versus electric field is displayed in Fig2-10. to realize the influence of different treatment on dielectric characteristics.
Among various post-treatments, the pure SCCO2 fluids SiOx film exhibits the most serious leakage current, inferentially due to its poor dielectric characteristics with numerous traps inside the SiOx film. The improvement of electrical characteristics is observed by using H2O vapor process, however, a high leakage current density still appears at larger applied voltages. It could be inferred reasonably dependent on the defect passivation efficiency. The most indicating that H2O vapor can passivate the traps (or defects) and alter dielectric properties of the low-temperature-deposited SiOx
film. After H2O vapor treatment, effective improvement of electrical characteristic is obtained by the 3000 psi-SCCO2 treatment, exhibiting the lowest leakage current density among all samples. Low leakage current density (less than 10-7 A/cm2). is kept constantly, even biased at an electric field of 3 MV/cm. The electrical performance agrees with FTIR analysis, in which 3000 psi-SCCO2 treatment modified SiOx
dielectrics even effectively.
2.3.2 Conduction Mechanism
There may be different conduction mechanisms in the insulator thin film, including Schottky-Richardson emission [23], Frenkel-Poole emission [23,24], Fowler-Nordheim tunneling [23,24], and trap assisted tunneling [25,26] illustrated in Fig 2-9. The Schottky-Richardson emission generated by the thermionic effect is caused by the electron transport across the potential energy barrier via field-assisted lowering at a metal-insulator interface. The leakage current governed by the Schottky-Richardson emission is as following:
A T exp E k T
JSR * 2 SR 12 SR B
where
0
12 find the slope of the leakage current equation.
The Frenkel-Poole emission is due to field-enhanced thermal excitation of trapped electrons in the insulator into the conduction band. The leakage current equation is:
k is the Boltzmann constant. We can find the slope of the leakage current equation. B
potential barrier. Tunneling is a quantum mechanical process similar to throwing a ball against a wall often results that the ball goes through the wall without damaging the wall or the ball. It also loses no energy during the tunnel event. The probability of this event happening, however, is extremely low, but an electron incident on a barrier typically several nm thick has a high probability of transmission. The Fowler-Nordheim tunneling current IFN is given by the expression [27]:
FN ox
where m is the effective electron mass in the oxide, m is the free electron mass, ox q is the electronic charge, and is the barrier height at the silicon-oxide interface B given in units of eV in the expression for BFN. is actually an effective barrier B height that take into account barrier height lowering and quantization of electrons at the semiconductor surface. Rearranging IFN formula gives by:
FN
FN ox through the oxide is pure Fowler-Nordheim conduction [27].In the trap assisted tunneling model, it is assumed that electrons first tunnel through the SiOX interfacial layer (direct-tunneling). Then, electrons tunnel through traps located below the conduction band of the high-k thin film and leak to substrate finally [25]. The equation of leakage current density is [26]:
ox
The plot of the nature log of leakage current density versus the square root of the applied electric field was observed. It is found that the leakage current density is linearly related to square root of the applied electric field. The linear variations of the current correspond either to Schottky-Richardson emission or to Frenkel-Poole
conduction mechanism. For trap states with coulomb potentials, the expression is virtually identical to that of the Schottky-Richardson emission. The barrier height, however, is the depth of the trap potential well, and the quantity FP is larger than in the case of Schottky-Richardson emission by a factor of 2.
Leakage conduction mechanism is also investigated to support the comments on the electrical improvement of SiOx film. Fig. 2-11(a) plots ln (J/E) versus reciprocal of electric field variation for the pure SCCO2 fluids treated SiOx film, and a schematic energy band diagram accounting for leakage transport mechanism shown in the inset.
A good linear fitting explains Fowler-Nordheim (F-N) tunneling [28] occurs in the electric fields higher than 1.2 MV/cm. Also, it is consistent with the electrical behavior of pure SCCO2 fluids treated SiOx film in Fig. 2-10 that leakage current density sharply increases, while gate bias voltage larger than 1.2 MV/cm. This could be attributed to the trap-assisted tunneling due to numerous traps inside the 150°C- pure SCCO2 fluids treated SiOx film [29]. For the 3000 psi-SCCO2 treated SiOx film, a plot of leakage current density versus the square root of the applied field (E1/2) gives a good representation of the leakage behavior at high electric fields, as shown in Fig. 2-11(b).
The leakage current density of the 3000 psi-SCCO2 treated SiOx is linearly related to the square root of the applied electric field, demonstrating Schottky-Richardson emission transport mechanism [30]. The Schottky-type conduction can be verified by comparing the theoretical value of
0
123
SR q 4
with the calculated one obtained from the slope of the experimental curve ln J versus E1/2 [31], where q is the electronic charge, the dielectric constant of free space, 0 is the high frequency relative dielectric constant. The Schottky emission generated by the thermionic effect is caused by electron transport across the potential energy barrier via field-assisted lowering at a metal-insulator interface, shown in the insert of Fig. 2-11(b), and independent of traps. From the slope of ln J versus E1/2, the calculated value of relative dielectric constant ( ) is 3.5, and which is close to the determined value of 3.8 in capacitance-voltage (C-V) measurement (referring to table 2-3). This also proves, for 3000spi-SCCO2 treated SiOx film, the conduction mechanism is really Schottky
emission, but not trap-dependent Poole-Frenkel emission [31]. Additionally, the evolution of conduction mechanisms from trap-assisted tunneling to Schottky emission can confirm these defects inside low-temperature-deposited SiOx film is minimized effectively by implementing the proposed SCCO2 technology. The leakage current densities of SiOx films after different treatments are shown as a function of applied positive gate bias voltage in Fig. 2-12, and the lower leakage current still could be acquired after 3000 psi-SCCO2 and H2O vapor treatment, especially treated with SCCO2 fluids. This could be attributed to the influence of traps in the interface between parasitical SiOx and Si wafer. Generally, in positive gate bias, the sources of electron are (1) the interface states, (2) defects in depletion region, (3) back electrode of substrate, [33] and the later two source are negligible due to the p-type signal-crystal Si wafer is used in this work. For pure SCCO2 fluids treated SiOx film, the great quantity of interface states still exist which generate electron-hole pair and lead to higher leakage current, as described in the inset of Fig. 2-12 After 3000 psi-SCCO2 treatment, the interface states were deactivated, hence the leakage current is reduced. The reduction of interface states would be proved in capacitance-voltage measurement. [36]
2.3.3 The capacitance-voltage (C-V) characteristics
The capacitance-voltage (C-V) characteristics are also generally used to judge the quality of dielectric films. Figure 2-13 shows capacitance-voltage characteristics of SiOx films after different treatment, measuring at 1M Hz with gate bias swing from negative voltage to positive voltage (forward) and from positive voltage to negative voltage (reverse). The slope of C-V curve in transient region, i.e. from Cmax to Cmin, is relative to the interface states, for example, the sharp slope indicates fewer defects exist in the interface between SiOx and Si wafer. In Fig. 2-13, the pure SCCO2 fluids -treated SiOx film presents the worst C-V curve and lower capacitance. This expresses the number of interface states exist and lead to the smooth C-V curve. Additionally,
the lower dielectric constant, as shown in table 2-3, could be referred to the influence of defects in SiOx film. With H2O vapor treatment, the sharper C-V curve and higher capacitance are obtained, and it could be attributed to the reduction of defects in SiOx
film and the interface. Furthermore, the best improvement is achieved by 3000 psi-SCCO2 treatment. This exhibits that the SCCO2 treatment possesses excellent ability to passivate the defects, including Si dangling bonds and interface states.
Besides, from Fig. 2-13, the shift of C-V curve under forward and reverse swing is also appears in pure SCCO2 fluids and H2O vapor-treated SiOx films. It is resulted from the trapped carrier in defects of SiOx films, and that is not expected for gate insulator of transistors. Under negative gate bias, the electric inject from Al gate into SiOx films and trapped by defects, leading to the larger gate bias is required for
Besides, from Fig. 2-13, the shift of C-V curve under forward and reverse swing is also appears in pure SCCO2 fluids and H2O vapor-treated SiOx films. It is resulted from the trapped carrier in defects of SiOx films, and that is not expected for gate insulator of transistors. Under negative gate bias, the electric inject from Al gate into SiOx films and trapped by defects, leading to the larger gate bias is required for