Chapter 1: Introductions
1.5 Thesis Organizations
In Chapter 1, the overview of vacuum microelectronics, basic principles of field emission theory, and research motivations are described.
In Chapter 2, we utilize the Co-Ti/Al (5nm/10nm) co-deposited catalyst compared to other kind of multilayer catalyst from our prior study. The improvements of reliability and uniformity in CNT FE-BLU, and increasing growth rate at low temperature are included to realize the properties and mechanism of Co-Ti/Al co-deposited catalyst thin film.
A simply lateral triode field emission device is fabricated in Chapter 3, we have utilized the cross-talk noise, which is one of the triode FE structure drawback, enhancing the uniformity of backlight unit.
Finally, the summary and conclusion are provided in chapter 4.
36
Chapter 2
Fabrication of Pillar-like CNT Field Emitters with Low Temperature Processes
2.1 Introduction
CNTs own some attractive physical and chemical characteristics, like high aspect ratio, adequate work function (~5eV), small tip radius of curvature, good chemical stability, strong mechanical strength, high conductivity, and electron emission properties [2.1-2.2].
However CNT-BLUs exist two crucial problems: one is inadequate lifetime which caused by poor reliability, and the other is bad uniformity which caused by screening-effect.
For reliability, two kinds of issues were observed: (1) abrupt decreases in emission current with increasing electric field and (2) a gradual degradation in emission current with high emission current density for a long period. Several reports indicated the weak adhesion on the interface of CNTs and substrate could cause an abrupt decrease in emission current resulting from a mechanical damage at high electric field [2.3]. Furthermore, high contact resistance between CNTs and substrate could result in a gradual degradation in emission current because of the Joule heat generated in a high resistive contact region [2.4]. Some methods have been reported for improving the adhesion or lowering contact resistance on the interface by post-treatment such as spin-on-glass (SOG) or polymethyl methacrylate (PMMA) coating, and zinc powder mixture [2.5-2.6]. Nevertheless, some of them might increase the complexity of processes and the cost of fabrication.
For uniformity issue, the screening-effect and non-uniform height of CNTs play the two most important roles. The simulation of the equipotential lines of the electrostatic field is shown in Fig.2-1 [1.58]. In order to avoid screening-effect, on some reported researches, such
37
as plasma post-treatment [2.7] and growing on AAO substrate [2.8], however these methods also increase the complexity and cost of processes.
Figure 2-1 Simulation of the equipotential lines of the electrostatic field for tubes of different distances between tubes. [1.58]
By thermal-CVD, the growth rate of CNTs at low temperature is slow. It has been known the size of catalyst nano-particles after pretreatment is critical importance of the CNTs growth, including the each CNT’s diameter, length, and density. Therefore controlling of the surface morphology of nano-sized catalyst particles is an essential prior to the CNTs growth.
According to Lindemann criterion, the melting point decreases as the catalyst particles sizes reduces as the Fig. 2-2 [2.9].
38
Figure 2-2 Lindemann criterion. [2.9]
And the nano-sized catalyst particles are more active compared with bulk catalyst metals due to surface effect [2.10]. The melting temperature of nano-particles is based on size-dependent cohesive energy by considering the surface effects. The melting temperature of nano-particles (Tmp) is linear to the reciprocal of the crystal size, i.e., Tmp= Tmb (1-C/D), where Tmb is the melting temperature of the corresponding bulk materials, D is the crystal size, and C is a material constant. Apparently, the proper determination of D is key issue. The smaller catalyst nano-particles with the lower melting temperature could be utilized to the CNTs growth at low temperature in thermal CVD to increase growth rates. From the AFM image as Fig.2-3 [2.11], the nano-particles of the co-deposited catalyst sample are more uniform and smaller than those of the conventional one. Accordingly, we applied the novel method to CNTs growth at low temperature for understanding the properties and mechanism of co-deposited catalyst.
39
Figure 2-3 AFM of the nano-particles after pretreatment
(a) the pure Fe catalyst layer, and (b) the Ti-Fe co-deposited catalyst layer. [2.11]
The schematic of a typical BLU is shown in Fig.2-4 [2.12] including light source, reflector, light guide, diffuser, and brightness enhancement film (BEF). The light source can be an incandescent light bulb, light emitting diodes(LED), cold cathode fluorescent lamp (CCFL), hot cathode fluorescent lamp (HCFL). All the backlights employ a diffuser and a BEF. The diffuser posited between the light source and the display panel is used to scatter the light for display uniformity. The BEF is used to enhance display brightness. The cost structure of materials for TFT-LCDs is described as Fig. 1-10 [1.57].
If we success to solve the reliability and uniformity problems, CNT-BLU will replace the traditional backlight system of LCD, it will be ensure to decrease amount of cost.
40
Figure 2-4 schematic of a typical backlight unit. [2.12]
Figure 1-10 cost structure of materials for TFT-LCDs of (a) 17inch LCD (b)32inch LCD. [1.57]
First of all, we used titanium as interlayer of the multilayer catalysts for CNT synthesis because it was found the most fitting one for CNT growth at low temperature. Then we fabricated pillar-like CNTs pattered structure on silicon substrate to improve the morphologies.
Our group has done some research on multilayer and co-deposited catalysts, but the catalyst are not the optimization for growing CNTs at low temperature. Finally, we accomplished some analysis for the above mentioned. The whole experimental procedures is shown below.
41
2.2 Experimental Procedures
2.2.1 Forward Arrangement
We have chosen 4 inch n-type silicon (100) wafer as our experimental substrate. After RCA clean and lithography processes, we defined several kinds of pillar-like circle patterns for CNT flied emission arrays which diameter is 6μm. The pillar-spacing is (1) 25 μm in experiment A, (2) 25 μm in experiment B, and (3) 3 to 30 μm in experiment C.
In our research, 1000Å Chromium layer was deposited by dual E-gun evaporation (JAPAN ULVAC EBX-10C) as the cathode between the substrate and catalyst. The catalyst in experiment A, we have chosen four kinds of different catalysts:
1. (Fe-Ti) 20Å iron and 30Å titanium co-deposited catalyst, 2. (Co-Ti) 20Å cobalt and 30Å titanium co-deposited catalyst,
3. (Co-Ti/Al) 20Å cobalt and 30Å titanium co-deposited on 100Å aluminum catalyst, 4. (Co/Ti/Al) multilayer catalyst formed of 20Å cobalt, 30Å and 100Å aluminum.
These catalysts were sequentially deposited on substrates by magnetron sputtering (Ion Tech Microvac 450CB) at the pressure of 7.6×10-2 Torr at room temperature. This sputtering system consisted of three sputtering source for different material targets and two power source for co-deposition, so our multilayer catalysts could be sequentially deposited without breaking the vacuum environment and co-deposited uniformly.
2.2.2 CNTs Synthesis
In our researches, an atmospheric pressure thermal chemical vapor deposition (T-CVD) system is used for CNTs synthesis, shown in Fig.2-5. It consists of a 2-inch-diameter horizontal quartz tube, an electric heating system, reaction gas supply, and related mass flow controllers.
42
(a)
(b)
Figure 2-5 (a) Photograph and (b) schematic picture of thermal CVD.
Firstly, Samples loaded into the quartz tube were heated to the predetermined temperatures (500℃ to700℃) in nitrogen flow (1000 sccm) to avoid catalyst be reacted during steps of heating. Secondary, before CNTs growing, hydrogen flow (50 sccm) was into
43
quartz tube about 5 minutes to reduce the catalyst metal to the metallic phase, then transforming into nano-particles. Thirdly, CNTs were grown at designated temperatures according to the optimal results of our group past researches, we had decided the flow rate of hydrogen is 10 sccm, 1000 sccm for nitrogen and 125 sccm for ethylene[2.13]. Last of all, samples were furnace-cooled to room temperature in nitrogen flow (3000 sccm). The schematic of process is shown below in Fig.2-6.
Figure 2-6 Process of CNTs synthesis (an example of CNTs growing 45 min at 550℃.
44
2.2.3 Analysis
The morphologies of pre-treatment catalyst were observed by atomic force microscopy (AFM), and the morphologies of CNTs’ samples were characterized by scanning electron microscopy (SEM), and we have used Hitachi S-4700I SEM in our researches. The finer internal structures of interface of CNTs and nano-sized catalytic materials were examined by high-resolution transmission electron microscopy (HRTEM), JEOL JEM-2000EX and X-ray energy dispersive spectroscopy (EDS) respectively.
Electric characteristics of CNTs field emission were measured with a parallel diode-type configuration in a high-vacuum chamber with the pressure of 5×10-6 Torr. A glass substrate coated with indium tin oxide (ITO) and P22 phosphor (ZnS: Cu, Al) was used as the anode plate, and the distance between the cathode and the anode plate was set to be 150 μm. The emitting area was variable, which was determined by pillar-spacing and pattern-area.
Anode voltages is sweep-type from 0 V to 1000 V which were applied at intervals of 10 V by a source measure unit (Keithley 237), as Fig. 2-7 for the verification of field emission characteristics while the cathode was biased at 0 V.
45
Figure 2-7 High vacuum measurement system.
46
2.3 Experimental Design
The scheme of the whole experimental procedures was shown below as Fig. 2-8.
Figure 2-8 The scheme of the whole experimental procedures.
47
2.3.1 Experiment A: Comparing Different Catalyst for CNT Growth
Constituent of catalyst is important to provide catalysis for CNTs’ growing, different materials and constituents give different reactivities. Here, four kinds of different catalysts, (Fe-Ti) 20Å iron and 30Å titanium co-deposited catalyst, (Co-Ti) 20Å cobalt and 30Å titanium co-deposited catalyst, (Co-Ti/Al) 20Å cobalt and 30Å titanium co-deposited on 100Å aluminum catalyst, (Co/Ti/Al) multilayer catalysts formed of 20Å cobalt, 30Å and 100Å aluminum were sequentially deposited on substrate by magnetron sputtering system. The total experimental process of profile was shown schematically in Fig. 2-9 on next page.
The procedures of pretreatment and CNTs growth was shown in Fig.2-6 before. The growing time is 60 minutes for each temperature, which vary from 500℃ to 700℃ with interval of 50℃. Finally, this method should be successful to find out the most appropriate constituent of catalyst for our predictive growing temperature.
48
Figure 2-9 Fabrication flow diagrams (a) ~ (k). (b) 100 nm Cr electrode deposited by E-gun, (f) four constituents kinds of catalyst by sputtering system, (j) pretreatment with H2 (50 sccm), and (k) CNTs growing under C2H4
atmosphere.
49
2.3.2 Experiment B: Effect of Growth Time
The growing time is a great part of cost for CNT-BLUs fabrication, because we need to keep high temperature during the CNTs growing duration. So the optimal growing time will obviously obtain both better field emission properties and lower cost we need.
In experiment B, the schematic of process is same as Fig. 2-6 except the growing time, and we designed five different durations at 550℃, that was 10 minutes, 30 minutes, 60 minutes, 90 minutes, and 120 minutes. The height of CNTs was predicted to increase as increasing of growing time. Following the F-N theory, we could easily know that higher pillar-like CNTs provide better field emission characters. However that character maybe not still be existed, because of too longer time of growth.
Lastly, we predicted we could optimize the growing time economically at high temperature in experiment B for low cost.
2.3.3 Experiment C: Optimization of Pillar Spacing
One of the critical drawbacks of CNT field emission backlight unit is non-uniform light region. So we were trying to overcome this drawback by using pillar-like pattern in order to avoid screening-effect.
In our researches, we chose several spacings between two pillar patterns, such as 3μm, 6μm, 9μm, 12μm, 15μm, 20μm, 25μm, and 30μm. The schematic of process is using the optimal growing time and predetermined temperature. Recipes were shown in 2.3.1 already and fixing these recipes except the spacing between pillars.
Finally, we predicted we could overcome the screening-effect and attain the most uniform lighting region, and find out the just spacing in experiment C.
50
2.4 Results and Discussion
2.4.1 Optimum Catalyst for CNTs Grown at Low Temperature (Experiment A)
Prior to CNTs growing, the catalyst is the first recipe we should focus on. According to our group’s researches, there were many advantages of CNT field emission characters by using novel co-deposition of catalyst, as Fig. 1-13.
(a)
(b)
Figure 1-13 Improvement of Luminescent Uniformity via Synthesizing the Carbon Nanotubes on an Fe–Ti Co-deposited Catalytic Layer. [1.62]
The most obvious character was the reliability increased quite substantially. We have guessed this phenomenon is caused by decreasing the diameter of nano-sized particles, so finer particle provide higher activity and lower melting temperature. As a result, we chose
Under 7.7 V/μm Under 7.7 V/μm
51
cautiously several kinds of catalyst compounds in order to obtain the just catalyst at predetermined temperature.
Fig. 2-10 shows the schematic profiles of thin film catalyst change into nano-sized particle during pretreatment, Fig. 2-10(a) illustrated two separations of two films and Fig.
2-10(b) illustrated co-deposited catalyst.
(a)
(b)
Figure 2-10 The schematic profiles of thin film catalyst change into nano-sized particle
52
during pretreatment (a) multilayer (b) Co-Ti co-deposited layer.
Figure 2-11 AFM images of different catalysts growing at different temperatures.
53
First, experiments with several catalysts of different constituents were only processing the pretreatment steps with hydrogen at 500, 550, 600, 650, and 700℃ separately for observing the reactive properties, roughness, and particle size. The AFM top view is the easiest method for making a comparison and it will clearly show the particle size and roughness. Following Fig. 2-11 is the whole AFM images in experiment A, the white pixel is standing for highest pick and the dark pixel is standing for the valley.
Next, we compared catalysts of tri-components with catalysts of bi-components, with or without aluminum between electrode and other catalyst layers. Subsequently, we could easily find out the differences. The fluctuation of catalyst surface showed simple and repeated nano-sized particle when pretreatment of bi-components catalyst without 10 nm Al buffer film, on the other hand, the fluctuation of catalyst surface was complex and additional curvature under those nano-sized particles when using the tri-components catalyst with 10 nm Al layer.
As this result, we knew that Al plays a role of providing an additional curvature for surface of catalyst and quite increasing roughness mean square (RMS). The advantage of increasing RMS is raising the density of catalyst particle on the same top view area, as Fig 2-12, and then we could obtain better density of CNTs of one pillar after the growing step.
As well as some drawbacks of CNT growth at low temperature are poor density of CNTs and weak adhesion between CNTs and substrate. We estimated that increasing CNTs density will not only improve the density of emission sites, but also solve problem of weak adhesion by increasing Van Der Waal force between CNTs.
(a) (b)
Figure 2-12 The catalyst after pretreatment (a) without Al buffer layer (b) with Al buffer layer.
54
Secondly, focusing on the Co-Ti/Al and Co/Ti/Al, the differences between them were methods of deposition, multilayer and co-deposited layer. Following the AFM [Fig.2-13(a)-(b)] and SEM [Fig.2-14(c)-(d)] images at 500℃, the phenomenon due to co-deposition different from multilayer was more obvious in SEM image, as shown below:
(a) (b)
(c) (d)
Figure 2-13 The top view at 500℃ (a) AFM of Co-Ti/Al, (b) AFM of Co/Ti/Al, (c) SEM of Co-Ti/Al, and (d) SEM of Co/Ti/Al.
Here, we assumed two mechanisms for phenomena of each method of deposition. For multilayer catalyst film, we explained that by surface energy of interface, the most effective factor is the melting point and difference of surface energy. But for co-deposited catalyst film, we took it as an alloy or a solid solution, it meant the constituent is uniform as same as other alloys. It precipitated when we pretreated at higher temperature, consequently explaining that by not only surface energy mechanisms but also nucleation and growth mechanisms.
55
(a)
(b)
Figure 2-14 The surface energy effected on interface reaction.
56
Surface energy of different atomic numbers are shown in Fig. 2-14(a) [2.14]. Focusing on three kinds of atoms “Fe, Co, and Ti” was utilized, we could easily get a unique character that those surface energy are nearly the same, about 1700~1800 erg/cm2. We expected to achieve the finest nano-particles of Co or Fe catalyst particles, and depending on the theory of surface energy, the most important rule is that everything of a system prefer to keep or diffuse to the lowest energy state after a long time diffusion. Following are three situations effected by difference of surface energy [as Fig. 2-14(b)]:
First, if the surface energy of buffer-component Ta we used is lower than Co (Co and Ta as an example), the lowest energy state will be larger Co particle and finer Ta, and almost Ta particles are inside of Co particle just for decreasing the total Ta surface area, so we may obtain larger and rougher catalyst particles with buffer-component of larger surface energy.
Second, if we choose a buffer-component with smaller surface energy, for example Co and Hf, the condition will be similar to the first situation. We could obtain the finer nano-sized catalyst particles by utilizing Hf buffer-component, but lots of Co catalyst particles were in the inside of larger Hf particles and could not react with carbon atom when CNTs growth.
Third, if the two elements with nearly same surface energy, as Co and Ti, the particles of each element will dispread uniformly to the similar sized nano-particles after hydrogen pretreatment. Depending on the past researches in our group, we were experimentally success demonstrating the Ti is the best buffer-component when we utilized Co or Fe as the catalyst element, the result was demonstrated our assumption, so we just used Ti as buffer-component in this study.
Another mechanism is nucleation and growth, which is the main factor of liquid phase transform to solid phase or solid solution phase precipitate particle when annealing as Fig.
2-15.
57
Figure 2-15 Liquid phase transform to solid phase or solid solution phase precipitate particle when annealing.
The co-deposited catalyst, as a solid solution was partial determined by nucleation and growth rate except the surface energy in our assumption. Based on this theory, there are two major factors affecting the nucleation rate:
(1) One is the instability of solid solution state as an over-saturated solution. So the more instability provides the higher driving force and faster rate of nucleation.
(2) Another is the diffusion rate of atoms into clusters, that means faster particle creation bring with the higher diffusion rate.
These two factors interact with each other, the scheme of the nucleation rate is Fig.2-16(a).
58
(a) (b)
Figure 2-16 The scheme of (a) the nucleation rate, and (b) the growth rate.
Figure 2-17 The nucleation rate of Co and Fe.
As Fig.2-16(a), the vertical axis was nucleation rate, the horizontal axis was temperature,
59
and Tm represented the melting temperature. The first curve represented increasing of nucleation rate as increasing of instability, the second curve represented increasing of nucleation rate as increasing of diffusion rate. We could simply realize there is a peak of the nucleation rate at a critical temperature.
Next, following the Fig. 2-16(b), the growth rate is simple comparatively. There was only factor of diffusion rate to affect the growth rate, therefore higher temperature brought higher diffusion rate, finally increase the growth rate.
Co and Fe catalyst were widely used for CNTs growing in our group’s past researches, and Ni was not a good catalyst element under hydrogen pretreatment and growing CNTs in thermal CVD at any temperature.
According to the phase diagram or atomic period table, the melting temperature of Co and Fe are separately about 1700 K and 1800K, and other characters of these two elements were quite similar. Thus we made a simulation of the nucleation rate of them which was shown in Fig. 2-17. Melting temperature was the main difference between Co and Fe, so that the curve of Co nucleation rate we plotted was a shift from the curve of nucleation rate of Fe.
In conclusion, we could assume surely the temperature of maximum nucleation rate of Co is low than maximum nucleation rate of Fe. By the way, focusing on the phase diagram as Fig.
2-18, we could find out the lowest temperature of α-Fe and α-Co generating which α-phase
2-18, we could find out the lowest temperature of α-Fe and α-Co generating which α-phase