Results and Discussion

2.4.1 The Reliability Improvements of Carbon Nanotubes for Field Emission Applications

The SEM micrographs of the CNTs in the conventional and the proposed samples were shown in Figs. 2-5(a) and 2-5(b) correspondingly. The roots of the CNTs for the proposed samples exhibited a little inserted (indicated by an arrow in Fig. 2-5(a)) but those for the conventional ones were only terminated on the surface (indicated by an arrow in Fig. 2-6(b)).

The transparent phenomenon implied a better conductive region around the roots. The inset of Fig. 2-5(b) displayed a SEM image of the proposed samples cleaved across the patterned region where a CNT with part of it plunged into the co-deposited metal layer was observed on the cleaved edge (marked by a circle). Therefore, the larger contact area and stronger adhesion for the proposed sample were depicted. The emission currents of both specimens were measured from 0 to 7.7 V/μm at 5×10^-6 torr and the curves of emission current density versus electric field (J-E curve) were shown in Fig. 2-6. For the conventional samples, an abrupt decrease of emission current in the first measurement was indicated by an arrow in Fig.

2-6(a). The decrease in emission current did not recover in the subsequently measurements and the curves of the next 4 measurements were almost the same. It can be attributed to that

part of the CNTs with poor adhesion were pulled off at high electric field during the first measurement and the remained CNTs with sufficient adhesion can sustain the force induced from electric field to provide smooth curves without abrupt decrease. On the other hand, for the proposed specimens, the curve of the first measurement was very close to the next 4 curves and no obvious abrupt current drop was observed as shown in Fig. 2-6(b). It resulted from that the enlarged contact area of the partially immersed structure of CNTs can provide sufficient adhesion to overcome the electric-field-induced force during measurements. Figs.

2-7(a) and 2-7(b) displayed the emission current density versus time for both the conventional and the proposed specimens tested at 7.7 V/μm for 3,600 sec. The emission current density of the conventional samples reduced from 60 mA/cm² to 20 mA/cm² after 700 sec and to 10 mA/cm² after 2,500 sec. However, the proposed samples exhibited a relative stable emission current density with 30 mA/cm² for 3,600 sec. Fig. 2-7(c) was the plot of the emission current density versus time for the conventional samples at 6.25 V/μm for 3,600 sec. The current density in the initial was about 30 mA/cm² as the same as the proposed samples at 7.7 V/μm, but unstable and reduced to 10 mA/cm² after 500 sec. It might be attributed to that the partially immersed structure with larger contact area can suppress the contact resistance and reduce the Joule heat generated in high resistive contact regions and prevent the CNTs from heat induced damages. For the first 400 sec in the Fig. 2-7(c), the current density was a little increasing because of the thermal emission. Fig. 2-8 showed the cross-sectional views of SEM for both samples before and after the stress for reliability analysis. Obviously, part of the CNTs in the conventional specimens disappeared after the time stress but the CNTs in the proposed ones had almost no change. It also manifested that the proposed samples with a partially immersed structure can prevent the CNTs from physical damages and improved the reliability of emitters.

For the CNTs formation mechanism has been proposed as the following steps: (i) carbon

source gas decomposition at the surface of the catalyst nanoparticles; (ii) formation of surface carbide in the reaction zone; (iii) carbon diffusion into the nanoparticle volume; (iv) carbon release, after some oversaturation at the catalyst nanoparticle.

In order to realize the growing process of CNTs for the proposed and conventional samples with respect to Ti-Fe codeposited thin film and pure Fe one as catalyst layer. The two mechanisms were put forward in this section. According to the calculated surface energy for period elements diagram, the surface energy of Fe was higher than Ti as shown in Fig. 2-10 [2.15]. As a result, for the conventional specimens, the ultra-thin Fe catalyst would tend to melt and agglomerate into nanoparticles during pretreatment due to different surface energy from the Ti buffer layer. And the nanoparticles were only laid on the surface of Ti buffer layer.

In contrast, for the proposed ones, the co-deposited Ti film would merge with the underneath Ti buffer layer and the Fe atoms simultaneously nucleated to form nanoparticles during pretreatment. So the nanoparticles in the proposed samples were partially immersed in the Ti metal. Accordingly, the synthesized CNTs for the conventional samples were only terminated on the surface of Ti metal but the CNTs for the proposed ones were partially immersed in the Ti metal. The schematic plots to depict such mechanisms for these two specimens were illustrated in Figs. 2-10(a) and 2-10(b). Therefore, this amazing structure of proposed sample provided better adhesion and lower contact resistance between the CNTs and the substrates. It could prevent the CNTs from heat induced destruction to improve the reliability of CNTs for the application as electron emitters.

2.4.2 The Improvements of Carbon Nanotube Growth Rate at Low Temperatures

The morphological images of CNTs taken by SEM were displayed in Fig. 2-11. From the comparison between the images of Fig. 2-11(a) and 2-11(b), the CNTs in the proposal samples exhibited a much longer length as compared with those in the conventional ones. The length

of conventional samples was about 437 nm, but that of the proposed ones was about 2.13μm.

It might be attributed to the suppression of coalescence of the Fe nanoparticles in the Fe-Ti codeposited film during the CNTs growth. It was remarkably observed by the SEM images of Fig. 2-12. Most of the catalytic particles in the proposal samples as Fig. 2-12(b) were smaller than those in the conventional ones as Fig. 2-12(a). For the conventional samples, the range of the catalyst particle size was from 30 nm to 170 nm, and the average diameter of them was 97.4 nm. For the proposed samples, the range of the catalyst particle size was from 50 nm to 100 nm, and the average diameter of them was 81.1 nm. It has been known that the smaller size of the catalyst nanoparticle would be sufficient for carbon diffusion and supersaturation of carbon atoms in the catalyst nanoparticle due to shorter volume diffusion path length. In addition, the cap lift-off to occur and the growth of the CNTs to start, the nanoparticles must have sufficiently great curvature (i.e., it must be small enough) so that the graphene layers of the cap are sufficiently strained so that cap liftoff, and addition of carbon in tubular form, is energetically favorable. In other words, CNTs nucleation and growth will only occur if the catalyst nanoparticles do not exceed a certain maximum size [2.16]. Therefore, the higher growth rate was exhibited in the proposal sample. As a result, the longer length of CNTs was obtained in the proposal one.

In addition, the less length variation of CNTs was also shown in Fig. 2-11(b) than those in Fig. 2-11(a). It might be due to more uniform catalytic particles in the proposal samples as Fig. 2-13(b) than those in the conventional ones as Fig. 2-13(a). The morphology of the catalytic particles weas shown in the cross section of the AFM images (Fig. 2-13).The Raman spectrum of the CNTs for the conventional and the proposed samples showed similar results, as shown in Fig. 2-14. It means that there was no obvious difference of the crystallinity in the CNTs for these two samples. It also implied that the Ti co-deposited with Fe might not participate in the growth of CNTs for the proposed samples. The TEM images were shown in

Fig. 2-15(a) and Fig. 2-15(b) with respect to the conventional and proposed samples. The layer-by-layer structure of graphite could be clearly observed. They could also demonstrate that the CNT in this TEM image was a multi-wall structure. Moreover, the composition of the catalytic nanoparticle was analyzed by EDS via the TEM instrument. The EDS spectrum in Fig. 2-16 showed that only Fe, C, and Cu existed in the results of the EDS analysis. The peaks of Cu were from the Cu mesh which was used to hold CNTs in the TEM system. The peak of C was from the CNT and the peaks of Fe were from the catalytic nanoparticle enclosed in the CNT. No signal of Ti was detected which indicated that the Fe would not form the alloy with Ti in this process. From the images of TEM and the analysis of EDS, they implied that the Fe played a critical role in the growth of CNTs and Ti seemed not involve in the formation of CNTs by Ti-Fe alloy. The plots of current density versus electrical field for the both samples were shown in Fig. 2-17. The turn-on field was defined as the field with the current density of 10 μA/cm². For the conventional samples, the current density was much smaller than the proposed ones, and it was much less than 10 μA/cm² at 6.25 V/μm. On the other hand, the current density of the proposed samples was 3.36 mA/cm² at 6.25 V/μm, and the turn on field was 4.44 V/μm.

To explain the smaller and more uniform catalytic nanoparticles in the proposal samples, the formation schemes of the nanoparticles for the conventional samples and the proposed ones were plotted as Figs. 2-18. For the conventional samples, the Fe film directly transferred to nanoparticles at 550 ℃ and subsequently the coalescence of these nanoparticles proceeded due to the difference of surface energy between Fe and Ti, as shown in Fig. 2-18(a).

The formation mechanism of catalytic nanoparticles might be that Fe atoms nucleated at 550

℃ to form nanoparticles and the codeposited Ti atoms merged with the underlaid Ti buffer layer. Ti atoms in the proposed samples played a role to disperse the Fe nanoparticles. Owing to the existence of the codepsited Ti, it regarded and suppressed the coalescence of the Fe

nanoparticles, as shown in Fig. 2-18(b). Hence, the CNTs in the proposal samples had more uniform distribution and higher growth rate at low temperature as compared with those in the conventional ones.