3.2 Simulations…
3.2.1 Devices Design and Simulations
Simulations were performed to investigate the beam spreading of emission electrons
with commercial software (SIMION-3D) using the finite element method. It was found that
an initial energy and divergence angles of e-beams at the moment of emission from CNTs
were 5 eV and -90°~+90°, respectively [3.6]. The thickness of silicon oxide layer formed by
plasma-enhanced chemical vapor deposition (PECVD) is 1 μm. In simulations, emission of
electrons is assumed on a flat surface of CNT emitters. The voltage between the cathode and
anode plates is applied with 1 kV with the spacing of 550 μm, while the gate voltage is
applied at 80 V.
The conventional device [3.7] has a surrounding gate so that electrons are emitted from
the peripheral area of the extraction gate. Fig. 3-1 shows the cross-section view of the device.
The spot size has strong relation to gate length. From Fig. 3-2, when the length of the
extraction gate (red area) is ranged from 20um to infinite, the simulations are clear that the
range of the electron beam from emission site (green area) is strongly related to the gate
length which can change the electric potential. The red lines are equipotential lines.
It very clear that the gate length correlate to the electric potential which can influence the
electron beam trajectories. So if there is a way to design a proper shape of the gate, the
focusing effect may improve by easier method. The whole idea is that utilized a
non-symmetric gate area to impact the trajectories of the electron beam. Fig. 3-3 shows the
emission sites (green area) are by the both side of extraction gate (red area). However, the
focusing effect is unsatisfied. The problem is that extraction gate area is too large so that it
creates too strong electric field. Therefore, if gate area can be designed properly, the focusing
effect can be achieved.
Figure 3-4 shows the novel self-focusing structure which designs to achieve the focusing
effect. The simulation (Fig. 3-5) reveals that is perfect impact on trajectories of the electron
beam when the extraction gate splits into two pieces. The self-focusing gate structure has a
symmetric extraction gate area, which consists of a pair of linear electrodes closed to the
emission region of CNTs. The spot size shrinks down to the 232um. It is astonishing that can
change the spot size so easily by this simple method.
Simulation results show that the self-focus structure has a spot size of 232 μm in length
on the anodic plate, while conventional one has 622 um in length. It’s clear that the novel
structure could effectively reduce the spot size on the anodic palate, therefore achieving a
better focusing effect.
3.3 Experimental Procedure
Fig. 3-4 illustrates the schematic diagrams of the top and cross-sectional views of the
CNT field emission devices self-focusing gate structures. The conventional gate structure
shown in Fig. 3-1 has a square area of CNT emitters within the gate aperture, that is, the
emitters were surrounded with the gate electrode, which is similar to the normal gate structure.
The fabrication processes are similar to previous work described in detail elsewhere [3.8]. The
fabrication procedures of the triode structure for CNT-FED were shown in Fig. 3-6. A (100)
n-type silicon wafer as the substrate was cleaned by RCA clean. As shown in Fig. 3-6(a),
2000A Cr as the cathode, 1 um SiO2, and 2000A Cr as the gate were deposited layer by layer
using the E Gun, Plasma Enhanced Chemical Vapor Deposition (PECVD), and E Gun,
respectively. Then there are two masks in the lithography process as Fig. 3-6(b). One whose
shape is stripe is to define the gate region and isolate the neighbor devices. The other one
whose shape is square is to define the catalyst metal deposition region. As described in Fig.
3-6(c), the gate and SiO2 were etched in the wet and dry etching (HDP-RIE), respectively.
With the previously patterned photoresist layer as the shadow mask, 100A Al, 30A Ti, and
20A Co were deposited on the patterned Cr cathode by Sputter (Fig. 3-6(d)). Finally, the Al
and catalyst layers on photoresist were removed by the lift-off method as presented in Fig.
3-6(e), and transferred into the thermal CVD chamber for CNT growth immediately (Fig.
3-6(f)).
CNTs were grown with a multilayer catalyst at atmospheric pressure by thermal CVD.
The multilayer catalyst (Co/Ti/Al) form of Al (10nm), Ti (3nm), and Co (2nm) were
sequentially deposited by magnetron sputtering. Samples with catalysts loaded into the quartz
tube were heated to the designate temperature of 500℃ in a nitrogen flow, and followed by a
pretreatment process with hydrogen gas of 50sccm. Then, nanotubes were synthesized with
reaction gases, ethylene and hydrogen, at flow rates of 10 and 125 sccm, respectively for 30
minutes. The whole growth recipe is the same as mentioned before (chapter 2).
3.4 Results and Discussion
Field emission devices with a novel self-focusing gate structure using CNTs as emitters
have been fabricated. Without additional focusing electrodes, the self-focusing gate structure
utilizes a pair of gate electrodes parallel with the vicinity of emitters, which results in an
asymmetric emission area as compared with the conventional gate structure. Therefore,
electrons emit from the emitters give rise to an overlapping region on the anode plate so that a
reduction of spot size has been achieved. According to the simulation results and luminescent
images, this self-focusing gate structure had a well controllability on the trajectory of
electrons, and therefore showed a smaller luminescent spot size than the conventional one.
Because of the overlapping of electron beams, the luminescent spot sizes could be remarkably
reduced to 232μm in x direction as compared with 622μm for the conventional gate structure
which had a serious issue of beam divergence.
Figure 2-23 shows a HRTEM image of nanotubes grown with the multilayer catalyst at
500℃, which reveals a closed tip filled with catalytic metal particles and a multiwalled
structure consisting of the wavy graphite sheets aligned parallel to the tube axis. The inner
and outer diameters are about 10 nm and 25 nm, respectively. The correlative Raman
spectrum of nanotubes shown in Fig. 2-15(a) indicates that the intensity of D-band
(1250~1450 cm-1) is larger than that of G-band (1550~1600 cm-1). It is well known that the
crystallinity of nanotubes synthesized at low temperatures is poorer than those grown at
higher temperatures due to the formation of vacancies, grain boundaries or other defects, and
furthermore the deposition of amorphous carbon in outer walls. It is noted that the fine
structure of nanotubes shown in Fig. 2-23 consists of an outer layer of amorphous carbon,
which is conjectured to be the factor giving rise to a high intensity of D-band.
The conventional device has a surrounding gate so that electrons are emitted from the
peripheral area of the extraction gate. In the contrary, the self-focusing gate structure has a
symmetric extraction gate area, which consists of a pair of linear electrodes closed to the
emission region of CNTs. The SEM micrograph of the self-focusing gate structure and
conventional one are shown in Fig. 3-7, which illustrates the top view of the device. Fig. 3-8(a)
is shown the zoom-in self-focusing structure and Fig. 3-8(b) is the cross-section view of the
CNT emitters adjacent to the gate electrode according to area circled in Fig. 3-8(a).
The field emission characteristics of devices with conventional and self-focusing gate
structures are measured in a high-vacuum chamber with the pressure of 5×10-6 Torr. The field
emission curve of current density versus extraction gate voltage is shown in Fig. 3-9, and the
inset represents the corresponding FN plot. Although conventional structure has a high
emission current density than novel one, there is no enormous difference between them.
The photo-luminescent images taken via CCD camera are shown in Fig. 3-10. The spot
sizes are consistent with the simulation results, indicating that the self-focusing structure has
good functionality in controlling the spot size of electron beams. The conventional gate
structure without focusing electrodes could not well confine the electron beam due to the
divergence of electron trajectories so as to give rise to a large spot (Fig. 3-10(a)) on the plate,
which would resulting in a serious cross-talk noise between pixels. Therefore, it is elucidated
the self-focusing structure without additional focusing electrodes could efficiently reduce the
spot size (Fig. 3-10(b)) on the anodic plate, thus alleviating the issue of the electron beam
divergence.
Chapter 4
Summary and Conclusions
4 Summary and Conclusions
The direct growth of CNTs on various substrates by thermal CVD at low temperatures
has been researched for the fabrication of field emission displays. Owing to the higher
throughput and better uniformity, thermal CVD is still the most attractive method for CNT
growth. Although using soda lime glass substrate indeed costs down, the melting point of 550
℃ restricts the processing temperature at which CNTs are synthesized by thermal CVD. The
growth conditions of the CNTs have been successfully optimized for thermal CVD. In order
to achieve an effective growth of CNTs, it is very important to control the optimum carbon
concentration in the catalyst by controlling the flow rates of ethylene, hydrogen, and nitrogen.
According to the experimental results, the appropriate choice of the flow rates of hydrogen,
nitrogen and ethylene are 10, 100, and 125 sccm, respectively. The optimum flow rate
condition is testified by SEM, Raman spectrum and CNTs’ field emission characteristics
included density and stress. The thermal CVD method has been used to investigate the effects
Optimum growth rates are likely to be achieved with specific flow rate of hydrogen, nitrogen,
and ethylene. The growth rate is less when only C2H4 is provided without the presence of
hydrogen. Excess amounts appear to result in slower growth rates and form the carbon films.
As a results, the role of hydrogen during catalyst pretreatment and stage of CNT growth is to
promote nano-particle agglomeration and activate the catalyst. The presence of nitrogen can
enhance the carbon dilution and diffuse in the catalyst surface. But, too much N2 gas caused
over-diluted. Ethylene is nothing but only carbon source.
A multilayer catalyst (Co/Ti/Al) can be employed to enhance the growth of CNTs at a
low temperature of 500℃, even at 370℃, in the atmospheric-pressure thermal CVD with
optimum growth condition is as mentioned above. The temperature-dependent growth rate in
the Arrhenius plot for the multilayer catalyst revealed an activation energy of 0.89eV, which is
much lower than that required for the conventional catalyst in thermal decomposition
(1.54eV). The fact that CNTs could be deposited at low temperatures with multilayer catalyst
was ascribe to the combination of well-distributed small catalytic nanoparticles due to the Al
supporting layer and the higher activity due to the Ti interlayer. CNTs grown at 500℃ with
the multilayer catalyst exhibited a high emission current density of 26.5mA/cm2 which
appears to be promising for their application to FEDs.
A novel self-focusing gate structure with CNT emitters has been shown to have shows
structure. The results of simulations and luminescent images clearly indicate that the
self-focusing gate structure employed a pair of gate electrodes close to the emitters could
cause an asymmetric emission area and the emitted electrons traveling through the spacing
between cathode and anode plates would give rise to an overlapping region on the anode plate.
Because of the overlapping of electron beams, the luminescent spot size could be reduced as
compared with conventional gate structure which has a serious issue of beam divergence. The
spot sizes of the conventional structures are improved from 622um to 440um which
correspond to gate lengths. In contrast, the spot size of the novel structure can shrink down to
232um. In addition to CNTs, the novel structure could be applied to all kinds of emitter
materials, such as ZnO rods, silicon tips or nanoparticles. As a result, the self-focusing gate
structure with a simple manufacturing process is potential for the applications in FEDs.
Tables
z Table 1.1 Comparison between vacuum microelectronics and solid-state electronics
Items Solid State
Microelectronics Vacuum Microelectronics
Current Density 104 – 105 (A/cm2) similar Turn-on Voltage 0.1 – 0.7 V 5 – 300 V
Structure solid/solid interface solid/vacuum inte rface Electron Transport in solid in vacuum
Electron Velocity 3×107 (cm/sec) 3×1010 (cm/sec) Flicker Noise due to interface due to emission Thermal & Short Noise comparable comparable Electron Energy < 0.3 eV a few to 1000 eV Cut-off Frequency < 20 GHz (Si) &
100 GHz (GaAs) < 100 – 1000 GHz
Powe r small – me dium medium – large
Radiation Hardness poor excellent
Temperature Effect -30 – 50 °C < 500 °C Fabrication & Materials well established (Si) &
fairly well (GaAs)
not well established
z Table 1.2 Some comparison of several kinds of flat panel displays
Types CRT OLED TFT-LCD PDP
Thin panel Poor Best Better Good
Large panel Good Good Good Best
View angle Better Best Good Better
Response
speed Better Best Good Better
lightness Best Better Better Better High
resolution Better Best Best Good
Low power
consumption Good Good Better Good
Life Best Good Better Better
Low cost Best Better Poor
Contrast Best Good Better Better Color
modulation Best Good Better Best
z Table 1.3 Some comparison of several kinds of field emission displays
z Table 2.1 Field emission characteristics with different ratio of H2/N2
z Table 2.2 Field emission characteristics with different flow rate of N2
H2 / N2
z Table 2.3 Field emission characteristics with different flow rate of C2H4
(E≈6.25 V/um) 0.58 mA/cm2 6.84 mA/cm2 17.44 mA/cm2 19.4 mA/cm2
125 138
Fig. 1-1 Energy diagrams of vacuum-metal boundary: (a) without external electric field;
and (b) with an external electric field.
Vacuum
Fig. 1-2 The figure is the applications of flat panel display
Fig. 1-3 The schematic diagram of (a) field emission arrays (FEAs), (b) conventional cathode ray tube (CRT)
Cell phone Mobile navigator PDA
Portable DVD Electronic book Desk
(b) (a)
Fig. 1-4 The SEM micrograph of (a) Spindt type triodes array, (b) Spindt type field (a)
(b)
(c) (a)
Fig. 1-5 The SEM images of (a) pyramidal structures formed by wet etching, (b) the silicon tips formed by SF6 plasma etching ,(c) silicon tips sharpened by thermal oxidation
(a)
(b)
(c)
Fig. 1-6 The FED products based on Spindt type field emitters, (a) motorola 5.6” color FED, (b) Pixtech 15” color FED, (c) Futaba 7” color FED and (d) Sony/Candescent 13.1” color FED
(a) (b)
(c) (d)
Fig. 1-7 This diagram is the field emission mechanism of BSD
Fig. 1-8 Structure of an MIM device
Fig. 1-9 (a) The SCE top view, (b) the field emission operation diagram of SCE
Fig. 1-10 The growth rate variation with temperature for thermal CVD and PECVD.
The data points for thermal CVD and high temperature PECVD are from previous data by Ducati et al.. The values for the activation energies were calculated from a linear interpolation of the slopes
(a) (b)
Fig. 1-11 The relationship between melting point and Co particle sizes.
Fig. 1-12 In-situ TEM images recorded from a region of capped Pt nanocrystals at various specimen temperatures. Surface diffusion or surface premelting of nano-size particles takes place when the temperature is raised above 500ْ
C.[2.41]
Fig. 1-13 The schematic diagram for the CNT growth using multilayered catalyst (a) Multilayer deposition (b) Catalyst particles formation after pretreatment (c) CNT growth.
Fig. 1-14 Examples of various structures (a) planar gate [3-1] (b) double-gate [3-2] (c) mesh gate [3-5]
(a(a))
(b(b))
(c(c))
Fig. 2-1 Schematic experimental procedures
Fig. 2-2 A (a) photo and (b) schematic picture of thermal CVD (a)
(b)
Applied Voltage (Va) Vacuum Chamber
Anode-to-Cathod Distance = 160μm
Ground Plate Emission Current (Ia)
Spacer
Anode Plate
Si Wafer
Fig. 2-4 Diode structure fabrication flow diagrams (a) 2000A Cr cathode deposition by Egun, (b) Lithography procedure (c) Multilayer catalysts deposition, (d) Lift photoresist off, (e) Pretreatment, and (f) CNT growth.
Fig. 2-6 Process of Experiment A with different flow of N2
Fig. 2-7 Process of Experiment A with different flow of C2H4
Fig. 2-8 SEM micrographs of samples with the multilayer catalyst [(a)-(b)] after CNTs deposition processes at different H2/N2 Ratio. The inset of each figure shows the corresponding magnification tilted view.
Fig. 2-9 (a) I-V plot with the different ratio of H2/N2 and (b) its F-N plot H2 / N2 = 50 / 950
H2 / N2 = 0 / 1000 H2 / N2 = 10 / 990
H2 / N2 = 20 / 980
( (dd))
((aa)) (b(b))
(c(c))
(a(a)) ((bb))
Fig. 2-10 (a)Raman spectrum with the different ratio of H2/N2 and (b) its ID/IG
(a(a))
(b(b))
N2 = 0
N2 = 500 N2 = 750
N2 = 250
(d(d))
N2 = 1500 N2 = 2000
( (hh)) (c(c))
(a(a)) (b(b))
(e(e)) (f(f))
(g(g))
N2 = 1250 NN22 = 1000 = 1000 N2= 1250
Fig. 2-11 SEM micrographs of samples with the multilayer catalyst [(a)-(i)] after CNTs deposition processes at different flow rate of N2. The inset of each figure shows the corresponding magnification tilted view.
Fig. 2-12 (a) I-V plot with the different flow rate of N2 and (b) its F-N plot N2 = 5000
(i(i))
(a(a)) (b(b))
Fig. 2-13 (a)Raman spectrum with the different flow rates of N2 and (b) its ID/IG
(a(a))
(b(b))
Fig. 2-14 SEM micrographs of samples with the multilayer catalyst [(a)-(g)] after CNTs deposition processes at different flow rate of C2H4. The inset of each figure shows the corresponding magnification tilted view.
C2H4 = 25 C2H4 = 50
C2H4 = 75 C2H4 = 100
C2H4 = 125 C2H4 = 138
(a(a)) (b(b))
(c(c)) (d(d))
(g(g)) (f(f))
Fig. 2-15 (a) Raman spectrum with the different flow rate of C2H4 and (b) its ID/IG
(a(a))
(b(b))
Fig. 2-16 (a) I-V plot with the different flow rate of C2N4 and (b) its F-N plot
Fig. 2-17 SEM micrographs of samples with Co catalyst [(a)-(c)] after CNTs deposition processes at different temperatures. The inset of each figure shows the corresponding magnification tilted view.
600 ℃
(b(b)) (a(a))
650 ℃
700 ℃
(a(a)) (b(b))
(c(c))
Fig. 2-18 SEM micrographs of samples with Co/Ti/Al [(a)-(g)] after CNTs deposition processes at different temperatures. The inset of each figure shows the corresponding magnification tilted view.
700 ℃ 650 ℃
600 ℃ 550 ℃
(d(d)) (c(c))
(f(f)) (e(e))
(h(h)) (g(g))
(a(a)) (b(b))
500 ℃ 450 ℃
400 ℃ 370 ℃
Fig. 2-19 The temperature-dependent growth rates of CNTs synthesized with (a) the conventional catalyst (Co) and with (b) the multilayer catalyst (Co/Ti/Al). The (b(b))
(a(a))
Fig. 2-20 SEM micrographs of samples with the bilayer catalyst (Co/ Ti) [(a)-(e)]
after CNTs deposition processes at different temperatures. The inset of each figure shows the corresponding magnification tilted view.
500 ℃ 550 ℃
600 ℃ 650 ℃
( (dd)) (
(cc))
700 ℃ (e(e))
(a(a)) (b(b))
Fig. 2-21 SEM micrographs of samples with the bilayer catalyst (Co/Al) [(a)-(e)]
after CNTs deposition processes at different temperatures. The inset of each figure shows the corresponding magnification tilted view.
500 ℃ 550 ℃
(d(d)) (c(c))
600 ℃ 600 ℃
700 ℃ 600 ℃
(e(e))
(b(b)) (a(a))
Fig. 2-22 The temperature-dependent growth rates of CNTs synthesized with (a) the bilayer catalyst (Co/Ti) and with (b) the bilayer catalyst (Co/Al). The
activation energies are calculated from the slope of the linear fit to the data.
(b(b)) (a(a))
Fig. 2-23 A TEM image of CNTs deposited at 400 to 700 ℃with the multilayer catalyst
Fig. 2-24 The corresponding EDS analysis of the catalytic nanoparticle in Fig. 2-23 (500℃)
400ْ C 450 ْ C 500 ْ C
700 ْ C 550 ْ C
Fig. 2-25 The Stress test for experiment A samples (a) with different ratio H2/N2, (b) with different flow rate of N2 (c) with different flow rate of C2H4
0 500 1000 1500 2000 2500 3000 3500
0.5
0 500 1000 1500 2000 2500 3000 3500
-1
0 500 1000 1500 2000 2500 3000 3500
0
Fig. 2-26 The phase diagram for Ti and C
Fig. 2-27 AFM diagrams (a) with Al 10nm (b) with Co 2nm/ Ti 3nm and (c) with Co 2nm/ Ti 3nm/ Al 10nm
(c(c)) (b(b)) (a(a))
TiO2 (464.9)
TiO2-Al2O3 (458.7) TiO2 (458.9) Al2TiO5 (459.1) TiC (460.1)
Fig. 2-28 XPS diagram with Co 2nm/ Ti 3nm/ Al 10nm grows at 500℃
Fig. 2-29 The SEM images of (a) Co 2nm, (b) Co 2nm/Ti 3nm, and (c) Co 2nm/Ti 3nm/
Al 10nm after 500℃ pretreatment for 10 minutes (a(a))
(b(b))
(c(c))
Fig. 3-1 (a) The top view of the conventional structure, and (b) its cross-section view of the place of dash line
(a(a))
(b(b))
Fig. 3-2 The top view of the conventional structure (left) and its simulation (right) (a) (a(a))
(b(b))
(c(c))
(d(d))
Fig. 3-3 (a) The top view of the non-symmetric structure (left) and its cross-sectional view (right) and (b) its simulation
um
(a(a))
(b(b))
Fig. 3-4 (a) The top view of the novel self-focusing structure, and (b) its cross-section view of the place of dash line
(a(a))
(b(b))
Fig. 3-5 The simulation of the novel self-focusing structure
Fig. 3-6 Triode structure fabrication flow diagrams (a) 2000A Cr cathode, 1um SiO2, and 2000A Cr gate deposition by Egun, PECVD, and Egun, respectively. (b) lithography procedure, (c) define the gate and spacer by wet and dry etching, respectively (d) multilayer catalysts deposition, (e) lift photoresist off, (f) pretreatment, and CNTs growth
Fig. 3-7 The SEM microgragh of the top view of the (a) self-focusing structure, and (b) conventional structure
(a(a))
(b(b))
Fig. 3-8 (a) It is shown the zoom-in self-focusing structure and (b) is the cross-section view of the CNT emitters adjacent to the gate electrode according to area circled in Fig. 3-8(a)
(a(a))
(b(b))
Fig. 3-9 I-V plots for self-focusing gate structure (red curve) and conventional (black curve). The insert is F-N plot
Fig. 3-10 The luminescent images of the conventional structure (a) and the novel self-focusing one (b). (The red rectangles are emission sites)
(a(a))
(b(b))
References
Chapter 1
[1.1] J.Bardeen and W. H. Brattain, Phys. Rev. 74,230(1948).
[1.2] R. N. Noyce, US Patent 2,981,877(1959).
[1.3] S. M. Sze, “Physics of semiconductor devices”, 2nd ed., John-Wiley & Sons pulisher, New York, p. 648, 1991.
[1.4] R. H. Fowler and L. W. Nordheim, “Electron emission in intense field,” Proc. R. SOC.
A229, p. 173, 1928.
[1.5] F. M. Charbonnier, ”Developing and using the field emitter as a high intensity electron source,” Appl. Surf. Sci., 94/95, pp. 26-43 (1996).
[1.6] K. R. Shoulders, Adv. Computers 2 (1961), pp. 135-293.
[1.7] C. A. Spindt, I. Brodie, L. Humpnrey, and E. R. Westerberg, “Electrical properties of thin-film field emission cathodes with molybdenum cones,” J. Appl. Phys., Vol. 47, pp. 5248-5250, 1976.
[1.8] R. N. Thomas and H. C. Nathanson,” Photosensitive field emission from silicon point arrays,” Appl. Phys. Lett., Vol. 21, pp. 384-386, 1972.
[1.9] G. H. Chen, R.Q. Cai, X.M. Song, and J.X. Deng,” Preparation and field electron emission of microcrystalline diamond deposited on a porous silicon
substrate ,”Materials Science and Engineering B107 (2004), pp. 233–236.
[1.10] Chia-Fu Chen and Hui-Chen Hsieh,” Emission current influence of gated structure and diamond emitter morphologies in triode-type field emission arrays ,”Diamond and Related Materials 9 (2000), pp. 1257–1262.
[1.11] W. Rivera, J. M. Perez, R. S. Ruoff, D. C. Lorents, R. Malhotra, S. Lim, Y. G. Rho, E.
characteristics of carbon nanotubes ,”J. Vac. Sci. Technol. B, Vol. 13, No. 2, Mar/Apr 1995.
[1.12] R. E. Burgess, H. Kroemer, and J. M. Honston, “Corrected value of Fowler-Norheim field emission function v(y) and s(y),” Phys. Rev., Vol. 1, No. 4, p. 515, May, 1953.
[1.13] R. B. Marcus, T. S. Ravi, T. Gmitter, H. H. Busta, J. T. Niccum, K. K. Chin, and D.
Liu, “Atomically sharp silicon and metal field emitters,” IEEE Trans. Electron Devices, Vol. 38, p. 2289, 1991.
[1.14] P. Vaudaine and R. Meyer, “Microtips fluorescent display,” IEEE IEDM’91, p. 197, 1991.
[1.15] C. Curtin, “The field emission display,” International Display Research Conference,
[1.15] C. Curtin, “The field emission display,” International Display Research Conference,