Thermal gravimetric analysis (TGA)

In a Thermo gravimetric analysis, the percent weight loss of a test sample is recorded while the sample is being heated at a uniform rate in an appropriate environment. The loss in weight over specific temperature ranges provides an indication of the composition of the sample, including volatiles and inert filler, as well as indications of thermal stability. Set the inert (usually N2) and oxidative (O2) gas flow rates to provide the appropriate environments for the test.

In this experiment, Shimadzu TGA-50 was implemented on 5 mg of the as-grown SWNTs. We used razor blade to remove the as-grown SWNTs from

the substrate and collect 5-mg SWNTs sample to place in the alumina cell. Set the initial weight reading to 100%, and then initiate the heating program with 10

°C/min ramp rate. The gas environment is an oxidative decomposition (air, 30 sccm/min).

Chapter IV

Results and discussions

4.1 Microstructures and XPS spectra of the H-plasma-pretreated catalysts 4.1.1 Effect of catalyst precursor thickness and buffer layer

Figure 4.1 illustrates the typical SEM morphologies of the pretreated catalyst precursor on silicon wafer without application of buffer layer and with various CoCrPtOx catalyst precursor thicknesses: (a) 10 nm, (b) 5 nm, (c) 3 nm, (d) 2 nm, (e) 1 nm and (f) 1 nm, respectively, except Fig. 4.1(f) is the catalyst precursor morphology with 10 nm AlON as buffer layer. It indicates that the average catalyst size after pretreatment is smaller for the smaller catalyst precursor thickness, as also shown in Table 4.1. By comparing the same catalyst precursor thickness but with and without buffer layer application in Table 4.1, the specimen with buffer layer gives rise to a smaller particle size.

It is believed that very fine, dense and uniformly distributed catalyst particles after H-plasma pre-treatment can give rise to growth of the vertically and well aligned tubes due to mutual restriction between tubes. In contrast, the scarce particle distribution may result in growth of spaghetti-like CNTs. To examine this possibility, top-view and cross-sectional HRTEM micrographs of the H-plasma pre-treated catalysts without buffer layer are shown in Figs. 4.2(a)

Fig. 4.1 Top-view SEM images of the pretreated catalyst precursor on silicon wafer without application of buffer layer and with various CoCrPtO_x catalyst precursor thicknesses: (a) 10 nm, (b) 5 nm, (c) 3 nm, (d) 2 nm, (e) 1 nm and (f) 1 nm, respectively, except Fig. 4.1(f) is the catalyst precursor morphology with 10 nm AlON as buffer layer.

(a) (d)

(b)

(f) (c)

(e)

Table 4.1 CoCrPtO_x film thickness versus H-plasma pretreated average particle size.

1 H-plasma pretreatment conditions: microwave power = 600 W, H2 = 100 sccm, working pressure = 30 Torr, process time = 10 min, temperature = 580 °C.

2 For Si substrate with 10 nm AlON buffer layer.

and 4.2(b), respectively. It shows uniform distributed nano-particles with diameters ranging from 3 nm to 4 nm with an average value of 3.5 nm, which is quite close to 3.2 nm from SEM examinations in Fig. 4.1, as also shown in Table 4.1. Effect of buffer layer on the H-plasma pretreated catalyst formation is demonstrated in Figs. 4.2(b) and 4.2(c), where the catalyst precursor-coated substrates have the same thickness, but without and with AlON buffer layer, respectively. The corresponding EDX spectra for the H-plasma-pretreated catalyst precursors in Figs. 4.2(b) and 4.2(c) are shown in Figs. 4.3(a) and 4.3(b), respectively, indicating existence of buffer layer after pretreatment. It appears that the particle sizes after H-plasma pretreatment for the substrate with buffer layer are ranging from 1 ~ 3 nm (average 2.4 nm), which is smaller than the

Specimen

Fig. 4.2 HRTEM micrographs: (a) Top-view and (b) cross-section of pretreated 1 nm catalyst precursor without buffer layer. (c) Cross-section of pretreated 1 nm catalyst precursor with buffer layer. (d) Schematic diagram of Fig. 4.2(c).

corresponding value (average 3.5 nm) in Fig. 4.2(b). Effect of buffer layer is essentially to minimize the nano particle aggregation during H-plasma pretreatment. This is concluded from the cross-sectional view of Figs. 4.2(b) and 4.2(c), where the nano-particles seem to be trapped in the valleys of the surface, as schematically shown in Fig. 4.2(d). In additional to a buffer layer in Fig.

4.2(c), both Figs. 4.2(b) and 4.2(c) indicate formation of SiO2 layer (~ 4 nm) (a)

(c)

(b)

(d)

after pretreatment, which is formed in the earlier stage of the pretreatment and may act as buffer layer effect in Fig. 4.2(b).

Fig. 4.3 EDX spectra of the H-plasma pretreated catalyst precursors on Si substrates: (a) 1 nm catalyst precursor without buffer layer and (b) with buffer layer.

4.1.2 Effect of catalyst precursor oxygen content

Effect of oxygen contents of CoCrPt films is displayed by SEM micrographs in Figs. 4.4(a) to 4.4(d) for the H-plasma pretreated catalyst precursors, where the catalyst precursors of 10 nm thickness in Figs. 4.4(a) and 4.4(c) were coated on the substrate under Ar + O2 atmosphere (Ar:O2 = 10:30 sccm/sccm); and under pure Ar atmosphere (40 sccm) for catalyst precursors in Figs. 4.4(b) and 4.4(d). The catalysts in Figs. 4.4(a) and 4.4(b) were H-plasma pretreated at 515°C for 8 min; and at 591°C for 10 min in Figs. 4.4(c) and 4.4(d).

(a) (b)

Fig. 4.4 Top-view SEM images of 10-nm-thick (a) oxidized (Ar:O2 = 10:30) and (b) non-oxidized catalyst precursors with H-plasma pretreatment at 515°C for 8 min; and 10-nm-thick (c) oxidized and (d) non-oxidized catalyst precursor with H-plasma pretreatment at 591°C for 10 min. Other conditions: microwave power 600 W, working pressure 30 Torr and H2 100 sccm.

These figures indicate that the catalyst precursors coated under Ar + O2

atmosphere give rise to a much smaller particle size after H-plasma pretreatment than that under pure Ar atmosphere. Effect of oxygen is basically to force formation of the loosely-packed metal oxides aggregates in the catalysts, where the aggregates may then be reduced under H-plasma to become smaller metal

(c)

(a) (b)

(d) (d)

particles. Higher oxygen content of 10 nm catalyst precursor was coated on the substrate under Ar + O2 atmosphere (Ar:O2 = 10:40 sccm/sccm) for smaller particles, but it shows no more effect.

4.1.3 XPS spectra

Figures 4.5(a) to (c) presents the results of the XPS of the H-plasma pretreated CoCrPtOx film formed by PVD, respectively, to study how catalyst precursor can form 2~3 nm nano-particles and prevent the agglomeration of catalysts as the buffer layer did in an earlier work [Delzeit-2001-368] [Zhong-2005-1558]. The spectra indicate that PtOx, CrOx and CoOx phases are formed simultaneously in the as-deposited film during reactive sputtering deposition by PVD. Furthermore, the PtOx associated with a peak at 74.05 eV (4f7/2) comprises PtO (73.8 eV) and PtO2 (74.6 eV), while CrOx (577.4 eV) consists of CrO2 (576.3 eV) and CrO3

(578.3 eV), respectively. The spectrum presented in Fig. 4.5(a) shows that the Pt peak is shifted by 1 eV by pretreatment, indicating that most of the PtOx atoms are reduced to Pt atoms in the H2 atmosphere, but a small quantity of PtOx

remains. Most interestingly, the Cr phase in Cr2O3 remains, suggesting that the nano-particles from catalyst precursor results mostly from the reduction of PtOx

or CoO. However, the nano-sized catalysts are driven mainly by PtOx as will be discussed in the following paragraph.

Fig. 4.5 XPS spectra of as-deposited CoCrPtO_x film (upper curve) by PVD and after H-plasma pre-treated CoCrPtO_x film (lower curve).

Recently, PtOx has been demonstrated to be an active layer for use in the next generation of optical storage media applications [Kim-2003-1701], from which, oxygen is released after prolonged laser heating to compress the adjacent layers to form pits as recording marks, which consist of oxygen bubble gas and fine Pt nano-particles. Therefore, the exploding phenomenon associated with the reduction of PtOx (PtO/PtO2) may cause the formation of very fine particles and such self-assembly behavior may depend on the process temperature. In this experiment, a pre-treatment temperature that is too low (below 500 ^oC) results in an incomplete reduction of an oxidized CoCrPt film. In this case, non-uniform large catalytic particles are formed, and only MWNTs can be synthesized. The oxidized CoCrPt film has a higher critical self-assembly temperature than PtOx

(~500 ^oC), which difference is determined by the difference between the compositions.

With respect to the role of Cr, Cr2O3 can inhibit the agglomeration of particles [Shaijmon-2005-192]. Hence, it is believed to play a role in suppressing the agglomeration of catalytic particles in the self-assembly process. The Co element has an essential role in the dissolubility and precipitation of carbon species in the SWNTs growth because the dissolubility of carbon in Cr2O3 and Pt is rather small. The earlier work demonstrates that the pure Co catalytic nano-particles can also be formed by the reduction of CoO that is prepared as a

chemical complex solution. However, the large sizes distribution and low density of the particles do not support SWNTs growth. The self-assembly mechanism that involves the exploding effect associated with the reduction of CoCrPtOx film enables the size and distribution of the catalysts to be manipulated to fabricate as-grown SWCNT with the desired morphology by controlling the composition, thickness and H-plasma pre-treatment temperature of the film. The effect of the composition of CoCrPt on the self-assembly of nano-particles must be discussed in the future.

4.1.4 Morphology differences among various catalyst precursors

Additionally, to verify more carefully whether the effect is mainly resulted from PtO2, the as-deposited 5-nm-think Co and CoCrOx film are prepared as comparison to perform the experiment under the same process condition. The results of H-plasma pretreated 5-nm-thick Co, CoCrOx and CoCrPtOx films are illustrated in Figs. 4.6(a), (b) and (c), and show the particle size order is Co, CoCrOx and CoCrPtOx. From comparing particle size of Co and CoCrOx, we confirm that Cr2O3 has effective capability to inhibit nano-particles agglomeration. Moreover, comparison of CoCrOx and CoCrPtOx reveals the reduction of PtOx (PtO/PtO2) may cause the formation of very fine particles.

Although CoCrOx can obtain smaller particles than Co, it is not a good candidate

Fig. 4.6 Top view SEM images of after H-plasma pretreated 5-nm-thick (a) Co, (b) CoCrO_x and (c) CoCrPtO_x films.

(a)

(b)

(c)

as catalyst to synthesize SWNTs. Because there are no SWNTs can be observed in our experiments.

4.2 Morphologies and Raman spectra of the carbon nanostructures 4.2.1 Morphologies and Raman spectra of carbon nanostructures

Figures 4.7(a) to 4.7(f) illustrate typical FESEM morphologies of the as-grown SWNTs on silicon wafer with different catalyst precursor thickness: (a) 10 nm, (b) 5 nm (c) 3 nm, (d) 2 nm, (e) 1 nm and (f) 1 nm with buffer layer. To confirm the types of CNTs precisely, Fig. 4.8 shows the Raman spectrum of as-grown SWNTs film, respectively. The RBM peaks in the Raman spectrum [Fig. 4.7] indicate the presence of SWNTs in Figs. 4.7(d), (e) and (f). Therefore, we demonstrate that vertically aligned SWNTs are successfully synthesized using CoCrPtOx catalyst precursor with 2 nm and 1 nm thickness. And very high IG/ID ratio (43/1) is obtained by 1-nm-thick catalyst precursor with 10-nm-thick AlON buffer layer.

Figures 4.9(a) and 4.9(b) show the TEM images of as-grown SWNTs with 1-nm-thick CoCrPtOx film and 10-nm-thich AlON buffer layer. The results indicate that the nanotubes are grown as SWNTs bundles with diameters of 2 to 3 nm, which resembles the diameter of catalyst particles observed by HRTEM.

Therefore, each SWCNT maybe grown from individual catalytic nano-particles

and the diameter of tubes is determined by sizes of the catalysts. Hardly any

Fig. 4.7 SEM images of the as-grown SWNTs on silicon wafer with different catalyst precursor thickness: (a) 10 nm, (b) 5 nm (c) 3 nm, (d) 2 nm, (e) 1 nm and (f) 1 nm with buffer layer.

(c) (f) (a)

(b)

(d)

(e)

1200 1300 1400 1500 1600 1700

100 200 300 400 1200 1300 1400 1500 1600 1700

258279

Fig. 4.8 Raman spectra of the as-grown SWNTs film on silicon wafer with different catalyst precursor thickness: (a) 10 nm, (b) 5 nm (c) 3 nm, (d) 2 nm and (e) 1 nm. The RBM peaks and very high intensity ratio of G-band / D-band (43:1) are found in the (f) 1-nm-thick catalyst precursor with buffer layer.

Fig. 4.9 HRTEM images of the typical as-grown SWNTs where the tubes are mainly bundle type and their diameters are ranged from 2~3 nm.

MWNT are founded by TEM and a very high IG/ID ratio (43/1) is obtained by Raman spectrum, which indicate the purification and high quality of SWNTs is good as obtained by arc-discharge which is high temperature precess.

The tilt-view of specimen presented in Figs. 4.10(a) and 4.10(b) display SWNTs film with an ultra-high density and a tube density that considerably exceeds that reported elsewhere [Zhong-2005-1558]. Figure 4.10(c) magnifies a portion of Fig. 4.10(b), to show the fabrication of bundles of aligned SWNTs. Figure 4.10(d) shows that vertically aligned SWNTs with a height of ~40 micrometer were successfully synthesized. Notably, the SWNTs film can be easily removed from substrate and the SWNTs film is very flexible and soft, likely the woolen blanket, as shown in Fig. 4.10(e). Figure 4.10(f) shows a fragment of Si substrate with catalyst precursor which well-aligned SWNTs are magically grown exhibits the powerful growth ability of CoCrPtOx catalyst precursor.

Vertically well-aligned SWNTs films with heights of ~50 micrometer are thus concluded to fabricate successfully under the condition. By contrast, entangled SWNTs rather than aligned tubes were found in the specimens that comprised a thicker as-deposited catalyst precursor film of over ~10 nm and MWNTs are easily observed by TEM. These results are consistent with that the catalyst precursor thickness must be control at ultra thin to enable SWNTs to be synthesized.

Fig. 4.10 Typical SEM micrographs of extremely dense and vertically aligned SWNTs film on silicon wafers: (a), (b) tilt-view, (c) is the high magnification image of (b), (d) cross-sectional view, (e) shows very flexible morphology and (f) a fragment of Si substrate, respectively.

(a)

(b)

(d)

(e)

(c) (f)

4.2.2 Effect of specimen stacking sequences on carbon nanostructures

The schematic diagram of specimen placement in MPCVD is shown in Fig 3.4, and different placement gets different result. Sample A obtains the highest IG/ID ratio of as-grown SWNTs, because of the relative higher temperature by directly plasma bombardment to backside. But in sample C, only short and entangled SWNTs can be observed, as shown in Fig. 4.11. This is probably caused by plasma etching to SWNTs during growth process. Sample D gets the highest SWNTs because of the rough backside surface of Si wafer which gives more space for SWNTs growth. Although the placement of specimen will affect the result of SWNTs growth, the degree is relative slight comparing with other conditions, such as catalyst precursor, gas ratio, etc.

Fig. 4.11 Top-view SEM image of short and entangled SWNTs of sample C.

4.2.3 Morphologies of the two-step grown carbon nanostructures

Two-step growth of SWNTs using 1^st-grown SWNTs removed substrate is attempted under the same H-plasma pretreatment and growth conditions of section 4.2.1. Figure 4.12 show the SEM image and Raman spectrum of two-step grown carbon nanostructure, and the results shows that well-aligned SWNTs are successfully grown again and indicate the base growth mechanism.

From the low magnification SEM top view, it is observed that vertical well-aligned SWNTs are successfully synthesized again, but randomly oriented cracks appear on all two-step SWNT samples, as shown in Fig. 4.12(a). In contrast, cracks were never observed on previous samples. The most probably reason for these cracks is the insufficient density of two-step SWNTs by poisonous catalyst which amorphous carbon atoms are not completely removed in 10 minutes H-plasma pretreatment time. Therefore, the Van der Waals attraction among SWNTs creates cracks by contraction when the samples are cooled from 600°C to room temperature. Hence, we extend the H-plasma pretreatment time to 15 minutes to eliminate the amorphous carbon atoms on the catalysts, and then vertical well-aligned SWNTs without cracks are synthesized.

The Raman spectrum of two-step SWNTs with 15 minutes H-plasma pretreatment time is shown in Fig. 4.12(b). It is of interest that the RBM peak becomes singular centralization at Raman shift of 281 cm^-1, indicates that the

100 200 300 400 1200 1300 1400 1500 1600 1700

diameter of two-step SWNTs are highly uniform. This is probably resulted from the uniform catalyst particle size by longer H-plasma pretreatment time. Thus, extremely dense and vertical well-aligned SWNTs (40~50 μm) are successfully re-grown and the average yield rate one cycle is ~0.2 mg/cm².

Fig. 4.12 (a) Top-view SEM image of re-grown SWNTs with insufficient pretreatment time. (b) Raman spectrum of re-grown SWNTs with suitable pretreatment time.

4.3 Effect of scanning local laser heating pretreatment on CNTs growth 4.3.1 Morphologies of the laser-pretreated catalyst precursors

Figure 4.13(a) shows the morphology of the 1-nm-thick catalyst precursor deposited PC substrate after local laser heating pretreatment. The laser heating region which rises above other area is about 1-μm-diameter size. Figure 4.13(b) is high magnification image of Fig. 4.13(a), shows there are many tiny seams on

(a) (b)

the catalyst surface, which is probably result from the exploding reduction of PtO2.The morphology of laser heating region after 10 minutes 350 °C H-plasma pretreatment is shown in Fig. 4.13(c). It is apparent that the catalyst precursor at pattern region becomes well-distributed nano-particles and other area only exist a small amount disorder and bigger particles. We firmly believe that these tiny seams increase the plasma etching area and many defects occur on their surface during the process of PtO2 exploding reduction which enhances the etching ability of H-plasma. Thus, the catalyst nano-particles are successfully synthesized at 350 °C by laser heating and then H-plasma pretreatment. AlON buffer layer was not employed here, because it was thoroughly damaged after local laser heating pretreatment.

Fig. 4.13 Top-view SEM images of catalyst precursor (a), (b) after laser ablation pretreatment and (c) after 10 min. 350 °C H-plasma pretreatment.

(c)

(a)

100 200 300 400 1200 1300 1400 1500 1600 1700

190 Si RBM

G band D band

Intensity (arb. units)

Raman shift (cm^-1)

4.3.2 Morphologies and Raman spectra of carbon nanostructures

Top-view SEM image and Raman spectrum of low temperature grown SWNTs at 373 °C are shown in Figs. 4.14(a) and (b). From these, we confirm that SWNTs are successfully synthesized at laser heating region. Although many outcomes must be improved like the entangled SWNTs, weak RBM peak and low IG/ID ratio, etc, the feasibility for selective growth of well-aligned SWNTs and for compatibility with IC processes at low temperature is expectable in the future. The lowest temperature to synthesize SWNTs is 373 °C and MWNTs is 262 °C. And these are never reached in the past research in the world.

Fig. 4.14 (a) Top-view SEM image and (b) Raman spectrum of low temperature grown SWNTs at 373 °C

(a) (b)

4.4 Growth mechanisms of SWNTs

4.4.1 High temperature growth mechanism

The EDX analysis is used to study the growth mechanism of as-grown vertically well-aligned SWNTs by novel CoCrPtOx catalyst precursor, and the result [Fig. 4.15] shows the catalysts existed in the SWNTs removed substrate, indicates the growth mechanism of as-SWNTs is base-growth mechanism. This is confirmed by the interestingly two-step growth method which 1^st-grown SWNTs is removed from substrate to grow SWNTs again under longer H-plasma pretreatment and the same growth conditions. The surface and body diffusion were both expected in the high temperature growth mechanism

[Hofmann-2005-036101].

The TEM results indicate that each SWNT is grown from individual catalyst nano-particle and tube diameter is determined by catalyst size. With respect to the fabrication of the extremely dense SWCNT film, highly dense very fine catalytic particles is proposed to be formed uniformly by H-plasma pre-treatment and aligned tubes are then grown vertically, constrained among the tubes. By contrast, if the distribution densities of fine catalytic particles are low, entangled SWCNT are formed.

4.4.2 Low temperature growth mechanism

Hofmann et al. proposed the surface diffusion: The low activation energy path for nanotube growth [Hofmann-2005-036101], find low activation energy of 0.4 eV for carbon surface diffusion on Ni and Co (111) planes, much lower than for bulk diffusion. In MPCVD, the plasma ionizes the gas and creates new and more reactive species, such as radicals, in the gas phase and/or the catalyst surface, as well as to cause a local surface heating that enables an efficient adsorption and diffusion of carbon atoms at low substrate temperatures. The plasma atmosphere may influence the detailed catalyst surface kinetics in many ways and may also supply carbon from the gas phase for SWNTs growth. This allows a reduction of the growth activation energy, as, e.g., atomic carbon can chemisorb directly on the catalyst. Therefore, we suggest that diffusion of carbon on the catalyst surface is the rate determining step at low temperatures. The effect of the plasma is to increase the dissociation of CH4, and etch any a-C, which may deposit on

Hofmann et al. proposed the surface diffusion: The low activation energy path for nanotube growth [Hofmann-2005-036101], find low activation energy of 0.4 eV for carbon surface diffusion on Ni and Co (111) planes, much lower than for bulk diffusion. In MPCVD, the plasma ionizes the gas and creates new and more reactive species, such as radicals, in the gas phase and/or the catalyst surface, as well as to cause a local surface heating that enables an efficient adsorption and diffusion of carbon atoms at low substrate temperatures. The plasma atmosphere may influence the detailed catalyst surface kinetics in many ways and may also supply carbon from the gas phase for SWNTs growth. This allows a reduction of the growth activation energy, as, e.g., atomic carbon can chemisorb directly on the catalyst. Therefore, we suggest that diffusion of carbon on the catalyst surface is the rate determining step at low temperatures. The effect of the plasma is to increase the dissociation of CH4, and etch any a-C, which may deposit on