Summary

By depositing a thin Ti capping layer on the hydrogen pretreated Fe nanoparticles, the diffusion of the carbon radicals generated in the thermal-CVD was resisted. The density of the CNTs was greatly reduced to a suitable density of 2×10⁷ cm^-2 with an optimal thickness of the Ti capping layer, 2 nm. The screening effect caused from the high density of the CNTs was effectively suppressed and therefore increase the local enhancement factor on the tips of the CNTs. By this way, the turn-on field was greatly reduced from 3.8 V/µm to 2.5 V/µm. On the other hand, from the results of the material analysis, the CNTs synthesized from the Fe nanoparticles capped by a thin Ti metal layer showed a well multiwalled structure with no obvious difference from those synthesized from the Fe nanoparticles without capping layer. The Ti capped on the nanoparticles seemed not involve in the growth of the CNTs.

Furthermore, the Ti capping layer would melt and hold the Fe nanoparticles after being heated. It enlarged the contact area between the CNTs and the substrates to enhance the adhesion and reduce the contact resistance. With strong adhesion between the CNTs and the substrates, the CNTs could sustain large electrostatic force induced from the applied electric field and therefore suppressed the abrupt decrease in the emission current density. Moreover, the reduction of contact resistance also ease the Joule heat generated with high current density passing through the high resistive regions. The high temperature due to the Joule heat could make the remained oxygen react with defective regions in the CNTs, cause the evaporation of the CNTs, or

degrade the interfacial structure between the CNTs and the substrates. Therefore, the abrupt decrease or gradual degradation of the emission current density were both suppressed by means of utilizing the Ti capping layer.

Chapter 3

The Improvements of Reliability and Uniformity for the CNTs Synthesized from the Fe-Ti Codeposited Catalyst

In this chapter, a novel method of preparing the catalyst was proposed for improving the reliability and uniformity of the CNTs synthesized in the thermal-CVD.

A partially immersed structure of the CNTs was achieved by using this novel method which enlarged the contact area to enhance the adhesion and reduce the contact resistance. The abrupt decrease and the gradual degradation of the emission current density were both suppressed by this way. Moreover, the coalescence between the Fe nanoparticles was also reduced to obtain the uniform nanoparticles in size and location distribution and, therefore, gained a uniform light emission.

3.1 Introduction

As discussed in the previous chapter, the reliability of the CNTs -based field-emission devices is one of the most critical keys that greatly determines the commercialization of the field-emission displays or the back-light units. Another critical issue that also plays a very important role for the applications of the CNTs in the field-emission displays or the back-light units is the uniformity of the CNTs in the field-emission devices. The uniformity of light emission in a CNTs-based field-emission displays and back-light units is the most difficult and emergent problems that needed to be solved. Especially for the field-emission displays, the

precise control of light intensity is required to obtain a display with high image quality.

However, the lack in the mechanisms of the growth of CNTs leads to large variations of the CNTs in length, diameter, direction, and structure during the synthesis processes. The length of the CNTs, especially, affects the field-emission characteristics most seriously. It is hard to fabricate CNTs with uniform length, diameter, and distribution no mater what synthesis method is used, arc discharge, laser ablation, or chemical vapor deposition. For the screening printing technologies, the CNTs with good crystallinity synthesized from arc discharge, laser ablation, or other high temperature processes are dealt with some physical and/or chemical post-treatments to get similar length and diameter but the uniformity of light intensity is still poor in general cases owing to the random distribution of the CNTs after the screen printing. For the CNTs grown on the substrates by chemical vapor deposition systems directly, it is still difficult to control the length, diameter, or morphologies of the CNTs to be uniform due to the diameter variation of the catalytic nanoparticles.

The non-uniformity in length, diameter, and location distributions of the CNTs could greatly affect the uniformity of the light intensity in the field-emission displays or back-light units. It has been reposted that the non-uniformity of the light emission is greatly attributed to the emission current density which is strongly affected by the aspect ratio of the CNTs[3.1]. Several methods, like post plasma treatment[3.2], catalyst pretreatment[3.3], surface treatment[3.4], critical bias applying[3.5], post laser treatment[3.6], post particle blasting[3.7] have been proposed to improve the uniformity of luminance in the CNTs-based field-emission displays or back-light units.

Unfortunately, the some of them cause structural damages by utilizing the post-treatments and the others require more complex processes that may increase the cost or reduce the throughput of products.

In this chapater, an Fe codeposited with Ti was utilized as a novel catalyst for the

synthesis of CNTs in the thermal-CVD. By this way, the coalescence between the Fe nanoparticles during the hydrogen pretreatment and the growth of the CNTs is suppressed and uniform Fe catalytic nanoparticles are observed. Therefore, uniform CNTs with small variation in length and uniform location distribution are gained after the synthesis process. A homogeneous light emission is observed on the glass coated with indium-tin-oxide (ITO) and green phosphor (P22) by using the CNTs synthesized from the Fe-Ti codeposited catalyst. Here the CNTs synthesized from pure Fe catalyst are called the conventional samples and those synthesized from the Fe-Ti codeposited catalyst are called the proposed samples. The percentages of the light emitting area for the proposed samples with the dimension of 1 cm×1 cm are improved from 12 % to 29 % at 500 V, from 41 % to 75 % at 600 V, and from 86 % to 100 % at 700 V as compared to the conventional samples with the same dimension. No extra mask and processes are required by using this novel Fe-Ti codeposited catalyst. Such a method exhibits its simplicity and merits which are very potential in the applications of the field-emission displays or back-light units to improve their luminescent uniformity.

3.2 Experimental Procedures 3.2.1 Sample Fabrication

The N-type silicon wafers (100) with low resistance (1~10 Ω/cm²) were utilized as the substrates in this experiment. An array of 5×5 square holes with 100 µm in length and 100 µm interspacing were patterned by the lithography system on a photoresist film. After the lithography processes, a 50-nm-thick Ti layer was deposited by the dual-electron-gun physical vapor deposition system as a buffer layer between the CNTs and the substrate. To compare with the proposed method, two kinds of samples were prepared as the conventional samples and the proposed

samples. For the conventional samples, a 5-nm-thick pure Fe film was deposited by the dual-electron-gun physical vapor deposition system as the catalyst on the Ti buffer layer. On the other hand, Fe and Ti were codeposited by the dual-electron-gun physical vapor deposition system as a novel catalyst on the Ti buffer layer with weight percentage of 64% for Fe and36% for Ti. Besides, the quantity of Fe was the same for the conventional samples. After the deposition of buffer layer and catalyst, the photoresist was remove by a lift-off process in acetone solution to leave the Ti buffer layer with the catalyst in the squared holes only. Both samples were loaded into the chamber of thermal-CVD to be pretreated in hydrogen ambient (H2/N2 = 400/600 s.c.c.m.) at 700 °C for 5 min and subsequently grew the CNTs in the same chamber with ethylene (C2H4/H2/N2 = 5/100/10 s.c.c.m.) at 700 °C for 15 min. To demonstrate the process steps more clearly, a flowchart was displayed in Figs. 3.1 schematically.

3.2.2 Material Analysis and Electrical Measurement

After the synthesis of CNTs, the images of the CNTs in both samples were taken by SEM and the crystallinity of the CNTs in the proposed samples was analyzed by TEM and Raman spectrum. An energy dispersive spectrometer was also applied to analyze the composition of the nanoparticles enclosed in the CNTs. The field-emission properties of the CNTs in both samples were measured in a vacuum chamber at 5×10^-6 torr, as described in Chapter 2. Both the conventional and the proposed sample were measured from 0 to 7.7 V/µm 5 times and stressed at 7.7 V/µm for 2,500 sec. After the stress, both samples were measured from 0 to 7.7 V/µm again to estimate if the curves were changed after being stressed at 7.7 V/µm for 2,500 sec.

For the uniformity issue, the morphologies of the catalytic nanoparticles after the hydrogen pretreatment were analyzed by an atomic force microscope and SEM for

both samples before the synthesis of the CNTs. The morphologies of the CNTs in both samples were also analyzed by the SEM. Furthermore, the samples were loaded into a vacuum chamber to measure the field emission current at about 5×10^-6 torr. The phosphor (P22) common used in the cathode-ray tube displays was deposited on a transparent conductive material, indium-tin-oxide (ITO), to be an anode electrode in the vacuum measurement system and the cathode voltage was applied to the silicon substrates. The measurement system was the same as that described in Chapter 2 (Fig.

2.2) schematically. To observe the images of luminescence, samples of a square with 1 cm² without pattern were also prepared by growing the CNTs with both the conventional and the proposed methods. The luminescent images of both samples were taken by a digital camera when the anode was applied at 500, 600, and 700 volts.

3.3 A Partially Immersed Structure of the Carbon Nanotubes for the Reliability Improvements

For both the conventional and the proposed samples, the micrographs of the CNTs synthesized in the thermal-CVD were taken by the SEM and displayed in Figs.

3.2. The micrograph for the conventional specimens in Fig. 3.2(a) displayed that the roots of the CNTs synthesized from the pure Fe catalyst seemed to terminate on the surface of the Ti buffer layer only. However, the micrograph of the proposed specimens in Fig. 3.2(b) showed that the CNTs synthesized from the Fe-Ti codeposited catalyst seemed to partially immerse into the codeposited metal layer. To check the partially immersed structure, the proposed specimens were cleaved across the patterned area and the micrograph along the cleaved edge was also taken by SEM.

As shown in Fig. 3.2(c), a CNT partially immersed into the codeposited metal layer

was observed along the cleaved edge and marked by a circle. Moreover, the microstructure of the CNTs synthesized from the Fe-Ti codeposited catalyst was analyzed by the TEM and the images were displayed in Figs. 3.3. The micrograph shown in Fig. 3.3(a) demonstrated a CNT with an enclosed nanoparticle. With higher resolution, a multiwalled structure of CNTs was clearly observed in Fig. 3.3(b).

According to the phase diagram of Fe-Ti shown in Fig. 3.4, the codeposited catalyst should form alloy FeTi during the growth process in the thermal-CVD at 700 °C[3.8].

To identify the composition of the enclosed nanoparticle in Fig. 3.3(a), an analysis of energy dispersive spectrum was also applied to the enclosed nanoparticles and the results were shown in Fig. 3.5. According to the results of the energy dispersive spectrum, only signals of Fe, Cu, and C were detected in the enclosed nanoparticles which implied that the Ti codeposited with the Fe in the catalyst seemed not to involve in the synthesis of the CNTs in the thermal-CVD. The signals of Cu originated from the Cu mesh used in the TEM as a supporter. Moreover, the Raman analysis was also applied to analysis the structure of the CNTs in the conventional samples and the proposed samples. As shown in Fig. 3.6, no obvious difference in the Raman analysis was observed which showed that the Ti codeposited with the Fe seemed not affect the crystallinity of the CNTs seriously. To illustrate the synthesis of the CNTs in both samples, the formation of the catalytic nanoparticles and the growth of the CNTs for both the conventional and the proposed samples were displayed in Fig. 3.7 (a) and 3.7 (b), correspondingly. As shown in Fig. 3.6(a), the Fe film formed Fe nanoparticles in the conventional samples resulting from that the surface energies of the Fe catalytic film was higher than the surface energies of the Ti buffer layer and, then, synthesized the CNTs which terminated on the surface of the Ti buffer layer only.

However, as shown in Fig. 3.7(b), the Fe atoms nucleated in the codeposited metal layer due to higher surface energy than Ti and the Ti in the codeposited layer merged

with the Ti buffer layer at the same time in proposed specimens. The surface energies of Fe and Ti were shown in Fig. 3.8 where the surface energy of Fe is 2.9 J/m² and of Ti is 2.6 J/m²[3.9-3.10]. In the proposed samples, the Fe nanoparticles were partially buried in the codeposited metal layer and the CNTs synthesized from those embedded Fe nanoparticles partially immersed into the metal layer were observed after the growth process in the thermal-CVD. Figures 3.9(a) and 3.9(b) showed the curves of the emission current density versus the applied electric field for the five measurements before being stressed and one measurement again after being stressed for both the conventional and the proposed samples, correspondingly. In Fig. 3.9(a), the 1st curve exhibited an abrupt decrease in the emission current density (indicated by an arrow) while the other curves from 2nd to 5th measurement were smooth and had no obvious difference to each other. The abrupt decrease in emission current density in the 1st curve might resulted from that part of CNTs with weak adhesion were pulled off from the substrates and the CNTs with strong adhesion kept the curves unchanged in the subsequently four measurements. In Fig. 3.9(b), no abrupt decrease in the emission current density was observed and the curves from 2nd to 5th measurements were almost the same with only slight difference from the 1st measurement. It resulted from that the partially immersed structure in the proposed samples provided a stronger adhesion between the CNTs and the substrates than those in the conventional samples. After the five measurements, both specimens were stressed under a electric field of 7.7 V/μm for 2,500 sec and the curves of emission current density versus operating time for both conventional and proposed specimens were plotted in Figs. 3.10(a) and 3.10(b) correspondingly. The emission current density of the conventional samples decreased from 60 mA/cm² to 20 mA/cm² after being stressed for 750 sec and the emission current density of the proposed ones maintained about 30 mA/cm² for 2,500 sec without obvious degradation. It may result

from that the reduction of the contact resistance between the CNTs and the substrates in proposed samples would suppress the Joule heating generated in the high resistive contact regions and, therefore, suppress the instability of emission current induced from high temperature. It showed that the CNTs synthesized from the Fe-Ti codeposited catalyst had the feature of better reliability than those synthesized from the pure Fe catalyst. After being stressed, the curves of emission current density versus applied electric field for both samples were measured again and plotted in Figs.

3.9. The emission current density was greatly reduced after being stressed for the conventional specimens as shown in Fig. 3.9(a). However, the curve for the proposed samples after being stressed was just slightly altered as shown in Fig. 3.9(b).

Moreover, the morphologies of the CNTs before and after being stressed for both conventional and proposed samples were taken by SEM and demonstrated in Fig. 3.10.

Obviously, the density of CNTs in the conventional samples was reduced after being stressed while the density of the CNTs in the proposed ones had no serious change in the morphology after being stressed. According to the curves in Figs. 3.9 and Figs.

3.10 and the micrographs in Fig. 3.11, the CNTs synthesized from the Fe-Ti codeposited catalyst could suppress both the abrupt decrease and the gradual degradation in the emission current density as compared to those synthesized from the pure Fe film. The improvements of the abrupt decrease in the emission current density for the proposed samples might result from that the enlarged contact area of the partially immersed structure could enhance the adhesion between the CNTs and the substrates to increase the mechanical strength of the emitters. Additionally, the enlarged area in the proposed samples also reduce the contact resistance between the CNTs and the substrates which could suppress the Joule heat generated with high current density passing through the high resistive regions and therefore ease the gradual degradation in the emission current density for a long operating time.

3.4 The Suppression of the Coalescence between Fe Nanoparticles for a Uniform Light Emission

The micrographs of the catalytic nanoparticles for the conventional and the proposed samples were taken by the SEM from top-view and displayed in Fig. 3.12 (a) and 3.11(b), respectively, after being pretreated in hydrogen ambient at 700 °C for 5 min in the thermal-CVD. As shown in Figs. 3.12, the catalytic nanoparticles in the conventional samples have larger diameter and diameter variation than those in the proposed samples. Additionally, the atomic force microscope analysis was also applied to the nanoparticles in both the conventional and the proposed samples and the results were shown in Figs. 3.13(a) and 3.13(b), correspondingly. A more uniform distribution of catalytic nanoparticles with smaller variation in diameter was also observed in the proposed samples, as shown in Fig. 3.13(b), as compared with the catalytic nanoparticles in the conventional samples, as shown in Fig. 3.13(a). After the synthesis of CNTs at 700 °C for 15 min, the morphological images of the CNTs were taken by the SEM and displayed in Figs. 3.13. With 45° viewing angle, the morphologies of the CNTs in both the conventional and the proposed samples were demonstrated in Fig. 3.14(a) and 3.14(b), respectively. Moreover, the cross-sections of the CNTs for both the conventional and the proposed samples were displayed in Fig.

3.14(c) and 3.14(d), correspondingly. According to the images in Fig. 3.14(a) and 3.14(b), the CNTs in the conventional samples showed a rough top plane and the proposed samples showed a flat top plane. From the comparison between the cross-sectional views of the CNTs of the conventional and the proposed samples in Figs. 3.14(c) and 3.14(d), the CNTs in the conventional samples also exhibited a

much larger length variation than the proposed ones. It had been reported that the growth rate of CNTs was dependent on the size of catalyst[3.11]. Therefore, the uniform CNTs in the proposed samples might contribute to the uniform distribution in the diameter of the catalytic nanoparticles after hydrogen pretreatment. Referring to the results of the material analysis in previous section, the Ti codeposited with the Fe seemed not involved in the synthesis of the CNTs or affected its crystallinity seriously.

To explain the uniform catalytic nanoparticles in the proposed sample, the formation schemes of the nanoparticles for the conventional samples and the proposed ones were plotted in Figs. 3.15(a) and 3.15(b), respectively. For the conventional samples, the Fe catalyst transferred from a film to nanoparticles directly at 700 °C due to the higher surface energy of Fe than Ti and subsequently the coalescence of these

To explain the uniform catalytic nanoparticles in the proposed sample, the formation schemes of the nanoparticles for the conventional samples and the proposed ones were plotted in Figs. 3.15(a) and 3.15(b), respectively. For the conventional samples, the Fe catalyst transferred from a film to nanoparticles directly at 700 °C due to the higher surface energy of Fe than Ti and subsequently the coalescence of these