3.5.1. EPD-diamond/CNT/Si substrates
It was found that the EPD treated CNT/Si substrates were also etching out fast under the LA-MPCVD plasma recipe of H
2:CH
4:CO
2=89:5:6. Therefore, in order to reduce the effect of oxidation, in the next set of experiments, CO
2 gas was removed from the process recipe.
Figure 9a and 9b show the Raman spectra of the EPD seeded CNT/Si substrates after 15 and 30 minutes of LA-MPCVD plasma treatment without CO
2 in the recipe of H
2:CH
4=97:3. It was found that even after 30 minutes of LA-MPCVD, CNT survived the plasma conditions (inset
Figure 9d). The corresponding Raman signal of the sp
3 peak was weaker than those of the graphitic D-band (
Figure 9a) after 15 minutes of LA-MPCVD, however, the Raman spectra started producing fluorescent signals after 30 minutes of LA-MPCVD as shown in
Figure 9b. There lies a definitive signal (not prominent enough due to background fluorescence) for diamond at 1330 cm
-1 which was taken from the sample in
Figure 9d. There are loosely scattered diamond nanoparticles, of around 80 nm diameter or less, all throughout the images in 9c and 9d. There is some occasional appearance of square or rectangular holes in the plasma treated EPD-diamond/CNT/Si substrate surface. Sizes of those holes vary from 250-350 nm. It is interesting to see that some of the holes were occupied by bright whitish nanoparticles - its dimensions was measured to be 250 nm. It can be assumed that dislodgment of such white particles resulted in the creation of those square or rectangular holes. The SEM images of figures 9c and 9d were taken from the intermediate regions of the individual CNTs (inset image of
Figure 9d). The sizes of the some of the CNTs are found to be as big as 25 µm. If the CNT lengths in
Figure 9d are compared with those in figures 5a (20 µm after15 minutes LA-MPCVD), it is concluded that even after 30 minutes of LA-MPCVD plasma treatment, EPD seeded CNT/Si survived diamond deposition conditions without complete etching, due to the absence of CO
2 in the recipe.
3.5.2. a. Effect of CH4
Section 3.1.1. and
Section 3.1.2. have already shown that the early periods of low temperature LA-MPCVD run could not produce fully covered, otherwise perfectly DND seeded silicon and sapphire substrates (even after 60 minutes of deposition). One of the reasons of the slow growth of diamond crystals, could have been rapid etching of the nanodiamond films by the CO
2 present in the plasma recipe. Therefore, it was further investigated to see the effect of removing CO
2 from the recipe during the LA-MPCVD nanodiamond growth. Researchers [
43] have shown that at the low CO
2 concentration (less than 0.5%), there is formation of SiC in competition with the growth of boron doped NCD during the LA-MPCVD process, which is due to the silicon contamination from the quartz tube. Moreover CO
2 (less than 1%) enhances the NCD growth rate (16 nm/hr) in continuous mode LA-MPCVD. But an increase in CO
2 gas over CH
4 gas percentages reduced [
11,
13] the growth rate (7 nm/hr). But there is no report so far about the effect of CH
4 percentages on the NCD growth pattern, specifically during the unstable periods of LA-MPCVD run, i.e., until the stable substrate temperature is achieved.
Figure 10a,
Figure 10b and
Figure 10c are the SEM images of the NCD crystals grown over the Si substrates with the increasing methane percentages (3% - 5%), only after first 15 minutes of LA-MPCVD. It is found that the individual DND seed crystals (
Figure 1a) grew in sizes in figures 10. The diamond nanoparticle density is calculated to be equal to 0.85×10
10/cm
2 in figures 11c for 5% CH
4 percentage over silicon substrates, to some extent less than the that calculated from the
Figure 1a for the DND seeds (2×10
10/cm
2). It demonstrates that during the early growth stage, the DND seeds only grow laterally in size, also with the disappearance of some seed crystals due to plasma etching. The Raman signal from the NCD film on Si substrate, grown at 3% CH
4, but after 30 minutes of LA-MPCVD (
Figure 10d), is shown to produce quite sharp sp
3 peak with FWHM of 22 cm
-1. There are also Raman peaks in the
Figure 10d corresponding to non-diamond carbon depositions at 1541 and 1600 cm
-1, respectively. It is observed that the dark contrasting Si substrate areas in the figures 10a, 10b and 10c are progressively diminished - indicating that with the increase in the CH
4 percentages, the DND seed crystals grow bigger in sizes, within a given LA-MPCVD time period.
If the NCD crystal densities are compared in between silicon and sapphire substrates after 15 minutes of LA-MPCVD growth at 4% CH
4 without CO
2 (
Figure 11a and 11b), it is found that there are 10 NCDs within an area of 200 nm × 200 nm square (
Figure 11a) on Si and there are about 20 NCDs on sapphire substrate (
Figure 11b). Hence it may be inferred that sapphire substrates are more effective in growing the DND seeds. Similar trend of double the number of NCD crystals present on the sapphire substrate in comparison to the silicon substrate after the 15 minutes of LA-MPCVD growth, at 5% CH
4 without CO
2 (
Figure 11c and
Figure 11d), is also observed.
Now if we compare the NCD crystal densities onto Si substrate with increasing CH
4 percentages, it is observed that there are more NCDs, 15 in number, within an area of 200 nm × 200 nm square (
Figure 11c). Similarly increase in CH
4 percentages from 4% to 5% on the sapphire substrate also increases the NCD crystal densities from 20 to 30 NCDs present within an area of 200 nm × 200 nm square (
Figure 11b and
Figure 11d) of the sapphire substrate after 15 minutes of LA-MPCVD growth. Thus, it can be concluded that the increasing methane percentages promoted LA-MPCVD growth of NCDs.
Figure 10.
LA-MPCVD growth of NCD crystals on DND seeded silicon substrates without CO2 in the recipe. SEM images showing early stages of CVD diamond growth at (a) 3%, (b) 4% and (c) 5% CH4 after 15 minutes of growth at 1500 W input average power. (d) Raman signal from such NCD crystals.
Figure 10.
LA-MPCVD growth of NCD crystals on DND seeded silicon substrates without CO2 in the recipe. SEM images showing early stages of CVD diamond growth at (a) 3%, (b) 4% and (c) 5% CH4 after 15 minutes of growth at 1500 W input average power. (d) Raman signal from such NCD crystals.
Figure 11.
Comparison of LA-MPCVD growth (1500 W, 5 cm, 15 minutes) of NCD crystals on DND seeded silicon and sapphire substrates without CO2 in the recipe. NCD crystal (samples # LA200605-1) SEM 50 kX magnification images on (a) Si and (b) sapphire substrates at 4% CH4. NCDs (samples # LA200605-2) on (c) Si and (d) sapphire substrates at 5% CH4.
Figure 11.
Comparison of LA-MPCVD growth (1500 W, 5 cm, 15 minutes) of NCD crystals on DND seeded silicon and sapphire substrates without CO2 in the recipe. NCD crystal (samples # LA200605-1) SEM 50 kX magnification images on (a) Si and (b) sapphire substrates at 4% CH4. NCDs (samples # LA200605-2) on (c) Si and (d) sapphire substrates at 5% CH4.
3.5.2. b. Effect of deposition time
Growth rates of linear antenna CVD is slower in comparison to conventional microwave plasma CVD growth of diamond crystals inside resonant cavity reactors. The typical growth rates for MPCVD is in the order of microns per hour [
1,
44], whereas inside LA-MPCVD, NCD grows by 5-20 nm per hour [
13] - depending on the presence of CO
2 percentages in the recipe and also on the use of pulse mode, which favours higher growth rates. Hence, it is understood that longer LA-MPCVD deposition time is required to fully cover the substrates NCD films. The available literature reports only about the long deposition periods (mostly 8, 15 and 20 hours of LA-MPCVD). There is little or no information about the growth of NCDs during the initial periods of LA-MPCVD, when the substrate temperature is not stabilised yet (
Figure 1a) and is continuously increasing by heating with plasma.
Figure 1b is the cooling-heating cycle for the samples # LA200605-3, with 11 minutes of short LA-MPCVD deposition period, shown with a sudden rise in substrate temperature (from 88
o to 270
oC) at 2800 W input average power and 5 cm stage to antenna distance. The experiment # LA200605-1 started from the room temperature with 4% CH
4 and the plasma heating lasted for 15 minutes, which continuously increased the substrate temperature up to 170
oC at 1500 W input average power and at 5 cm stage to antenna distance. Thereafter, the reactor was let to cool down below 140
oC (temperature-safety-limit of the reactor) before it was to re-open to unload and reload the next set of 4-substrates. In this way, the starting temperature for the next set of samples # LA200605-2 became 70
oC (not restarting from the room temperature) and again with 15 minutes of heating with 5% CH
4 in the hydrogen plasma, it could heat up to a little bit of higher substrate temperature of 200
oC at 1500 W input average power and 5 cm stage to antenna distance. Again, the experiment was stopped and the reactor was allowed to cool down to 88
oC (shown in
Figure 1b – depending on the time one takes to load and unload samples) before restarting # LA200605-3 for 11 minutes of plasma heating up to 270
oC, but at 2800 W power and 5 cm distance. The important point to be noticed is that the heating rates were, 9.9
oC/min for # LA200605-1, 8.6
oC/min for # LA200605-2, and 16.5
oC/min for # LA200605-3. The third LA-MPCVD heating rate was higher because of the very high input MW power level of 2800 W.
Figure 12a and
Figure 12c compare the diamond film growth behaviour on silicon substrates after 15 min. (1500 W) and 11min. (2800 W) of LA-MPCVD heating with 5% CH
4 in the hydrogen plasma recipe, respectively. There are almost equal numbers (about 45 diamond nanoparticles inside 500 nm × 500 nm square) of whitish bright NCD particles scattered on the dark contrasting silicon substrate in both the images. Although, there was a 4-minute reduction in deposition time with concomitant increase in MW input power by 1300 W, the
Figure 12c shows somewhat sharper white spots of NCDs on Si, which may be due to its wider substrate temperature range of 88
o-270
oC. On the other hand, sapphire substrate has more numbers (approximately 70 diamond nanoparticles inside 500 nm × 500 nm square) of NCDs scattered in the
Figure 12b. Moreover, there are also signs of nanodiamond agglomeration in
Figure 12b. It may be concluded that sapphire substrate favoured more lateral growth of DND seed crystals than silicon substrate. It may be attribute to a tendency for silicon to form carbide which reduces the formation of diamond nanoparticles. The number of NCD particles present inside 200 nm × 200 nm square area is about 20 in the
Figure 12d, for the LA-MPCVD growth on sapphire substrate, for 11 minutes with 5% CH
4 at 2800 W power and at 5 cm distance - which is almost equal in number of the NCD particle density found in
Figure 11b for 15 minutes of LA-MPCVD with 4% CH
4 at 5 cm distance. The number of diamond nanoparticles was 30 in
Figure 11d inside 200 nm × 200 nm square area for 15 minutes of LA-MPCVD with 5% CH
4 at 5 cm distance. So, this decrease in NCD density number (30 to 20) with identical LA-MPCVD recipe is mainly due to the decrease in deposition time from 15 to 11 minutes. It is important to note that although the MW input power was much higher (2800 W) - leading to higher substrate temperatures (88
o-270
oC), but it could not become effective in increasing the diamond nanoparticle density at shorter LA-MPCVD periods of 11 minutes.
Now, if the LA-MPCVD was allowed to proceed from 15 minutes to 30 minutes at 3% CH
4 in the process recipe without CO
2, with simultaneous lowering of the stage from 5 to 6.5 cm distance away from the quartz tube antenna, it was found (inside 500 nm × 500 nm squares) that the NCD crystal sizes grow bigger in sizes with concomitant touch with each other to form some kind of agglomeration (
Figure 13a and
Figure 13c) on silicon substrate surfaces. On the other hand, the NCD crystals are found already to be much bigger on sapphire substrates (
Figure 13b and
Figure 13d) under identical LA-MPCVD processing conditions than on the silicon substrates, which is again due to tendency of silicon to form carbide favourably over nanodiamond formation. There are about 15 NCD crystals/agglomerates inside 200 nm × 200 nm squares in
Figure 13b in comparison to approximately 10 NCD crystals/agglomerates inside 200 nm × 200 nm squares in
Figure 13d. The NCD sizes vary from as small as 25 nm individual crystals to as big an agglomeration size as 100 nm in
Figure 13d, whereas, the smallest NCD in
Figure 13b is about 20 nm and the biggest agglomeration is found to be as big as 50-60 nm. Therefore, it may be concluded that longer deposition time allowed the NCD to grow in the lateral direction in covering more the underlying sapphire substrate, gradually with time. The NCD crystals are found to be 40-45 nm in size (
Figure 13c) on silicon substrate after 30 minutes of LACVD with 3% methane in hydrogen plasma, occasionally touching each other, whereas, in
Figure 13a (15 minutes) they remained isolated from each other with smaller (30 – 35 nm) diamond nanoparticle sizes.