Catalysts are critical for carbon nanotubes growth. In catalyst for carbon nanotubes research, much attention was paid to the catalyst composition and the stabilization of catalyst [98-100, 112]. However, it is of great interesting to understand the role of chemical components in catalyst, the formation of catalyst nanoparticles and the CNTs growth on catalyst particles.
The main focus of this chapter is the understanding of the role of multi-components present in the catalyst for the growth of single-wall carbon nanotubes by chemical vapor deposition (CVD). We have chosen to evaluate the catalyst activity of the heterogeneous catalyst by varying the chemicals contents in the catalyst solution-based preparation. In the production of carbon nanotubes, catalysts play a distinct role since they are used to determine the type (multi-walled or single-walled) and the amount of nanotubes.
It is demonstrated that combining Scanning Electron Microscopy (SEM), Transmission Electron Microscopy (TEM) and Raman spectroscopy, X-ray diffraction (XRD), X-ray
photoelectron spectroscopy (XPS) and micro balance allows an evidence of the role of chemical components present in catalyst and optimization of crystallinity and morphology of catalyst.
3.1 Synthesis of nano-structured catalysts
In our experiments, a mixture of fumed alumina, iron (III) nitrate nonahydrate and bis (acetylacetonato) – dioxomolybdenum (VI) was chosen to be the catalyst to produce SWNT as shown in chapter 2. An impregnation method is used to prepare this catalyst. This method is used in order to have a homogeneous distribution of catalyst particles on the substrate on which carbon nanotubes would be grown.
The catalyst is prepared by impregnation process by mixing fumed alumina nanoparticles, iron (III) nitrate nonahydrate and bis (acetylacetonato) – dioxomoybdenum (VI) salts in methanol solvant. The catalytic suspension is ready to use after 30 minutes sonication.
According to the usage, we can prepare catalysts in two ways (fig.2.19):
Layer of catalyst on the Si/SiO2 substrate to study the role of chemical components and the CNT growth: this corresponds to branch (1) from fig. 2.19
Powder of catalyst to study the catalysis process: branch (2), branch (3) and branch (4)
For branches 2 and 3, we used a heat treatment process under argon and argon + hydrogen gas respectively, at 900°C to create catalytic powder. Furthermore, we have to analyze these catalysts powders to determine components which may be appearing in catalysts after CVD process. The catalyst can be spread evenly distributed over the whole sample or only in specific areas (fig. 3.1):
- A liquid based catalyst can be spread on the whole sample by spin coating or droplet method.
- For localised catalyst, a pattern of equidistant squares is written using e-beam lithography and the plasma etching step is replaced by spreading the liquid catalyst.
After this step some catalysts are dispersed on top of the PMMA and directly on the surface of substrate. The catalyst on the PMMA can be removed together with the PMMA, leaving behind only the patterned catalyst on the substrate (lift-off method).
Figure 3.1 Scanning electron microscopy images of Catalyst for the CNT synthesis.
(a). Catalyst on the whole surface. (b). The patterned catalyst.
All materials used in experiments are research grade: iron (III) nitrate nonahydrate, and bis(acetylacetonato) - dioxomolybdenum (VI) from Sigma Aldrich Chemicals, oxide C alumina nanoparticles from Degussa Inc. The alumina has a specific surface area of 100 m2/g, and the diameter of the particle is 13nm. The high-purity methane and hydrogen are supplied from Air Product.
3.2 The CNT growth
The CNT synthesis process is carried out in the EasyTube system (ETS) using the prepared catalyst. The scheme is presented in fig.3.2 below.
Figure 3.2 Schematic of a chemical vapor deposition system (ETS) for CNTs synthesis
H2 CH4
Ar
Furnace
Furnace
sample Quartz tube
Controller gas flow
Gas inlet
Gas outlet
For the carbon nanotube growth, a flow of 1000 sccm CH4, 250 sccm H2 is introduced into the chamber instead of Argon which is used during the heating step. In our experiments some different temperatures were experimented in the CVD process for the carbon nanotubes growth: 700°C, 750°C, 800°C, 850°C, and 900°C. Carbon nanotubes growth time was adjusted between 3 to 30 minutes.
3.3 Study the role of chemical components in the catalyst
To study the role of chemical components present in catalyst for the growth of single-wall carbon nanotubes, a series of samples (following branch 1) were prepared with various catalyst compositions (table. 3.1). In our experiment, we used the suggested catalyst of First Nano Inc as the standard catalyst [101]. This catalyst is suitable for SWCNT growth in the EasyTube system.
Table 3.1 Name and chemicals components of layer catalyst on substrate Name of catalyst Chemicals components
T1
Standard catalyst
Iron (III) nitrate nonahydrate Aluminum oxide C
Bis(acetylacetonato)-dioxomolybdenum(VI) TFe+Mo
Catalyst without alumina
Iron (III) nitrate nonahydrate
Bis(acetylacetonato)-dioxomolybdenum (VI) TMo
Catalyst without iron
Aluminum oxide C
Bis(acetylacetonato)-dioxomolybdenum (VI) TFe
Catalyst without molybdenum
Iron (III) nitrate nonahydrate Aluminum oxide C
The catalysts in table 3.1 and their products are characterized by SEM, TEM, Raman, XPS, and microbalance. We tried to analyze this kind of catalyst with XRD, however, we were obtained no signal of chemical components or new composition in the layer catalyst sample.
The reason of this problem is not well understood. This may be because the size of catalyst nanoparticles or the thickness of the layer catalyst was very thin (few ten of nanometers). So, the corresponding catalysts powders were prepared to be characterized with XPS and XRD
before and after thermal treatment at 900°C, following the branch (3) preparation mode (see table 3.2).
Table 3.2 Name and chemicals components of powder catalyst products Name of catalyst Chemicals component
T1p
Standard catalyst
Iron (III) nitrate nonahydrate Aluminum oxide C
Bis(acetylacetonato)-dioxomolybdenum(VI) TFe,p Iron (III) nitrate nonahydrate
Aluminum oxide C TMo,p Aluminum oxide C
Bis(acetylacetonato)-dioxomolybdenum (VI)
p: powder catalyst
3.3.1 SEM images of different chemical components in catalysts after CVD
The influence of chemical components in the catalyst on the nanotube growth has been studied systematically. After CNTs synthesis process with the Easy Tubes System, SEM (Geremi LEO 1530) was used to observe the surface of samples.
(a) (b)
(c) (d)
Figure 3.3 SEM images of products after CNTs synthesis by CVD process (a). Standard catalyst (Al, Fe, Mo). (b). Catalyst without Al2O3.
(c). Catalyst without Iron and (d). Catalyst without Molybdenum
SEM images of different catalysts exhibit an apparent morphological variation. Fig. 3.3a and 3.3c indicate that carbon nanotubes appear on the samples (using standard catalyst and catalyst without iron) during CNT synthesis. Whereas, in fig. 3.3b and 3.3d, we conclude that nanotubes are not present and didn’t appear during the synthesis. Especially, in fig. 3.3b, where large particles sizes of catalyst have been created after CVD process.
From fig. 3.3, one can observe that the alumina affects the catalytic particles size. Catalyst without alumina exhibits large particles size, a few ten of nanometers (fig. 3.3b). The catalytic samples in the presence of iron compound can grow CNT while the use of catalyst without iron compound results in the absence of CNT.
- At this stage, the role of molybdenum in this experimentation remains unclear. Catalyst with molybdenum oxide and alumina can’t grow carbon nanotubes in the conditions explored in this study. So, the presence of molybdenum compound in catalyst doesn’t catalyze the CNTs growth step. We will discuss the role of molybdenum compound in the next section.
From a comparison between all SEM images we deduce that the first role of the alumina in powder is to support the catalyst particles (iron and molybdenum particles). With the presence of alumina in catalyst, we obtained the catalyst nanoparticles after CVD process so, alumina prevents the small particles of the metal particles to agglomerate. From these experiments, we also deduce that iron is an active component to grow the carbon nanotube but molybdenum is not one (comparing between fig. 3.3c and 3.3d).
3.3.2 Raman spectra of various chemical components in catalyst
Making a Raman spectroscopy comparison between the standard catalyst with the different catalyst (MoO2(C5H7O2)2/Al2O3 system – TMo and Fe(NO3)3/Al2O3 system – TFe), the nature of carbon products can be determined. In this section, the Raman spectroscopy is only used to determine the appearance of SWNTs in our process.
The Raman experiment conditions for CNTs samples are presented below:
- Excitation ⇒ Argon 514.5nm - Laser power ⇒ 100mW - Power on sample ⇒ 1mW
- Mode ⇒ microscope objectif X100 - Configuration ⇒ Triple
- Time of integration⇒ 90s, 2 accumulations
Under laser illumination, it is necessary to limit the power of the laser, in order not to burn the nanotubes and avoid shifts of peaks.
Two principal areas of interest are:
- The low frequency area, between 100 and 300 cm-1, where we find the radial breathing modes of the nanotubes (RBM).
- The high frequency area, between 1400 and 1800 cm-1, where the tangential modes (TM) are located.
The diameter (d) is determined by measuring the RBM frequency and applying the formula:
νRBM = 224/d (nm) [27, 29-30].
Symbols of catalytic samples have been explained in table 3.1 above.
100 150 200 250 300 350 0
200 400 600 800 1000 1200 1400 1600
Intensity (a.u.)
Frequency (cm-1)
TFeLF TMoLF T1LF
1200 1300 1400 1500 1600
0 500 1000 1500 2000 2500 3000 3500 4000
Intensity (a.u.)
Frequency (cm-1) TFeHF
TMoHF T1HF
Figure 3.4 Raman spectra of CNTs samples (a). RBM peaks. (b). D-peak and G-peak
(a)
(b)
The various samples show different typical Raman spectra in fig 3.4, each taken from resonant nanotubes on the Si/SiO2 substrate. The Raman spectra of two samples TFe and T1 show that carbon nanotubes are present. These CNTs are SWNTs:
- The presence of some RBM peaks is observed in the range 130-300 cm-1
- The nanotube tangential graphitic G – band modes at 1560-1600cm-1 and disorder D-band modes at 1320-1380 cm-1 are also found in these results.
The Raman spectrum of sample TMo in this experiment reveals that the spectra doesn’t show RBM peak in the low frequency range neither G-peak, D-peak in the high frequency range. Thus, the sample with a catalyst composed of alumina and molybdenum oxide leads only to amorphous carbon. So, we deduce as for the SEM analysis that molybdenum compound is not active to grow CNTs in our process. To complete this result, we used the microbalance results to calculate and compare the yield of carbon products.
3.3.3 Yields of carbon products after synthesis CNT process
The weights of substrates, catalysts and carbon products have been measured by micro balance (the accuracy of measurement is 100 ng) at CEA-INAC laboratory. With these measurements, we determine the quantity of carbon products after the CNTs synthesis process. Then, we compared these values to fix the “best” chemical component ratios in the catalyst. With this “best” ratio, we can obtain the optimal amount of carbon products.
The procedure to determine the various weights of catalyst and carbon products are presented below. All of values are the average of three measurements on same samples.
For substrate
Silicium/silica substrates are stored in the air, so impurities can be adsorbed around them.
When we use these substrates in our experiments, they liberate impurities. The lost of weight of this substrate is leads to an error for the subsequent measurements. So the weight of these impurities must be determined to exclude this type of error.
The weight of substrates samples were measured before and after thermal treatment at 900°C in Ar and H2.
The thermal treatment process of substrates and catalysts can be show in the figure 3.5:
Figure 3.5 Thermal treatment process with ETS.
The impurities weight was calculated as shown below:
Mimp = Msub,b - Msub,a (Eq. 3.1)
The following equation defines the percentage of substrate’s lost weight after thermal treatment.
100 100
- 1
%M
, , ,
r = ×
=
b sub
a sub b
sub ipm
M x M
M
M (Eq. 3.2)
where: Msub,b ,Msub,a are weights of substrate before and after thermal treatment, respectively (determined by microbalance).
Mimp is the weight of impurities adsorbed on substrate.
%M is the real percentage of substrate’s weight which we used in r
experiments.
This value can be used to calculate the real-used weight of substrate and to expel error in later calculation.
The weight lost of catalyst due to thermal treatment
Chemical components in our catalyst are iron (III) nitrate nonahydrate, aluminum oxide C and bis(acetylacetonato)- dioxomolybdenum (VI). New components are created at high temperature. So, the weight of catalyst is lost during the thermal treatment process due to the formation of gases according to these reactions:
2Fe(NO3)3 Fe2O3 + NxOy↑ MoO2(C5H7O2)2 MoO3 + CO↑ + H2↑
Ar
H2
Ar
Room temperature
900 oC
Here, we use catalysts on the Si/SiO2 substrates as samples for the same thermal treatment process (figure. 3.5).
In this part, the micro balance provided three values:
- The weight of SiO2/Si substrates before thermal treatment (Msub, b TT)
- The weight of catalyst + substrate before and after thermal treatment (Mcat + sub, bTT, Mcat + sub, aTT)
The weight of catalyst before treatment (Mcat, b TT):
Mcat, bTT = Mcat + sub, bTT - Msub,bTT (Eq. 3.3)
As discussed above, we can use eq. 3.2 to calculate the intrinsic value of substrate and the weight of catalyst.
The remaining weight of catalyst after thermal treatment (Mcat, a TT)
Mcat, aTT = Mcat + sub, aTT - Msub,bTTx(%Mr) (Eq. 3.4)
The percentage of used catalyst before and after thermal treatment process:
%Mcat, aTT = 100
,
, x
M M
bTT cat
aTT
cat (Eq. 3.5)
where:
Mcat, bTT, Mcat, aTT are the weights of catalysts before and after thermal treatment.
Carbon products yield:
As the initial weight of catalyst and substrate introduced into the reactor are known, we measure the weight increase of our samples (substrate + catalyst + carbon products) after reaction. Samples are prepared according to process branch 1 of fig. 2.19. The weight of samples is measured by micro balance.
The weight of carbon products:
Mcar, p SP = Mcar+ cat + sub, a SP - Mcat, a SP - Msub, a Sp (Eq. 3.6) = Mcar+ cat + sub, a SP - Mcat,b SPx (%Mcat, aTT) - Msub, b SP x(%Mr)
= Mcar+ cat + sub, a SP – (Mcat+ sub, b SP – Msub,b SP) x (%Mcat, aTT) - Msub, b SP x(%Mr) The carbon product yield:
%MCar, p SP = 100
,
, x
M M
aSP cat
pSP
car (Eq. 3.7)
where: SP: Synthesis Process
Mcar, p SP is the weight of carbon products.
Mcat, a SP is the weight of catalyst after synthesis process
Mcar+ cat + sub, a SP is weight of samples after synthesis process (provided by microbalance)
Mcat+ sub, b SP is the weight of samples before synthesis process (provided by microbalance)
Msub, b SP is the weight of substrate (provided by micro balance)
T Fe T Mo T1
0 5 10 15 20 25 30 35 40 45 50
samples
Yield of synthesis CNT process (%)
T Fe T Mo T1
Figure 3.6 The yield of carbon products using different catalysts To understand the role of molybdenum compound, we compare the yield of carbon products for different catalysts in table 3.1. The results obtained in these experiments are consistent with the effect of bis (acetylacetonato)-dioxomolybdenum (VI) on catalyst described in section 1.
Fig. 3.6 demonstrates that the MoO2(C5H7O2)2/Al2O3 (TMo) system provides ~14 wt.%
yield. A slight increase of the yield is observed for the system Fe(NO3)3/Al2O3 (TFe)
compared to MoO2(C5H7O2)2/Al2O3 catalyst. It was mentioned above that iron appeared as the active agent for CNT growth. But, the highest catalytic activity is obtained for the mixture of iron nitrate, bis (acetylacetonato)-dioxomolybdenum (VI) and alumina (case T1).
Combined with the Raman and SEM results, we suggest that:
The produced carbon products on the MoO2(C5H7O2)2/Al2O3 system were almost amorphous carbon.
With Fe(NO3)3/Al2O3 and MoO2(C5H7O2)2/Fe(NO3)3/Al2O3 systems, the product are high quality bundle and individual carbon nanotubes.
So, molybdenum compound appear as the promoter for increased/improved production of carbon nanotubes.
3.3.4 Determination of crystallography structures in catalytic system by X-Ray Diffraction (XRD)
XRD method can determine the presence of crystalline structures of the various chemical components in catalyst, so we use this method to define the components in our catalysts.
The catalysts powders are prepared by the method presented in fig. 2.19 (chapter 2), branch (2), branch (3) and branch (4). Afterwards, they are deposited on a glass substrate using a mixing of ethyl acetate and glue. They are analysed by XRD using the copper Kα radiation.
Powder TFe,p
The first analysis result (fig. 3.7) confirms that the rhomboedrical Fe2O3 compound can be identified thanks to its diffraction peaks. It seems that there is no other compound (the presence of little peaks of low intensity is possible but not evident when zooming).
The bump at low 2θ angles is probably due to the signal of the glass substrate.
00-033-0664 (*) - Fe2O3 - Rhombo.H.axes - a 5.03560 - b 5.03560 - c 13.7489 sum of T_Fe - File: T_Fe_Add_Scans.raw - Type: 2Th/Th locked - Start: 20.00
coups/s
0 10 20 30 40 50 60 70
2 theta
20 30 40 50 60 70 80 90 100
Figure 3.7 XRD spectra of TFe,p sample Powder T1p (branch 3):
The analysis is not easy because the peaks are broad (presence of several compounds or low grain size) and the background noise remains high: nevertheless, the diffraction peaks (fig.3.8) lead to:
00-029-0063 (N) - gamma-Al2O3 - Cubic - a 8.00324 - b 8.00324 - c 8.00324 - 01-082-0582 (C) - (Fe0.807Al0.193)(Al1.807Fe0.193)O4 - Cubic - a 8.15010 - b 8 sum of T1 - File: T1_Add_Scans.raw - Type: 2Th/Th locked - Start: 20.000 ° - E
coups/s
0 10 20 30 40 50 60 70
2 theta
20 30 40 50 60 70 80 90 100
Figure 3.8 XRD spectra of T1 sample
- probably a cubic structure of (Fe0.807Al0.193)(Al1.807Fe0.093)O4 type - Possibly a cubic structure of Al2O3 type
The bump at low 2Θ angles is again probably due to the signal of the glass substrate (because of few quantity of powder)
The formation of the new composition in the catalyst after thermal treatment at 900oC is evidenced by the interaction between alumina support and active catalyst iron oxides. This interaction is very strong because it is created by chemical bonds. It must be noticed that a compound of Al2MoO4 type (either monoclinic or orthorhombic with very fine grain size) can also explain this bump. However, XRD results did not reveal any signal of molybdenum ions.
We strong believed that the slow amount and the very small particles of molybdenum are the reasons of this phenomenon.
Some XRD spectra of different catalyst powders which are prepared by the method presented in fig. 2.19 can be found in annex. 2.
3.3.5 Determination of the chemical compound formula in catalyst by X-ray Photoelectron Spectroscopy (XPS)
The powder catalysts were investigated by XRD as presented in sections 3.3.4. With XRD analysis, we can’t find the appearance of molybdenum compounds (the peaks large and low intensity). So, we use XPS method to determine the features of molybdenum and to confirm the results of XRD analysis.
Some experiments are performed on the layer catalyst, which is prepared by the branch (1), on Si without thermal silica oxide to prevent the effect of substrate.
Analyses have been realized on a XPS SSI spectrometer, using Kα Al X-ray source. The flood gun has been activated in order to control charge effects which are induced by the insulating nature of the surface of the catalyst powder.
Figure 3.9 The XPS spectrum of T1 sample.
(a). Complete XPS spectrum. (b). Peaks of iron.
(c). Peaks of molybdenum. (d). Peak of aluminium
Survey spectrum presented in fig.3.9 shows that all elements are detected: Al, Mo, Fe, O and C. Fig. 3.9b shows that Fe2p peak is detected, though it has been necessary for the
(b)
(c) (d)
(e)
(b)
(d)
(c)
adaptation of the acquisition parameters (e.g. resolution, time) in order to get a satisfactory result. The binding energy of the 712 eV peak, corresponds to Fe2O3. The Mo3d spectrum in fig. 3.9c shows a single doublet characteristic of a unique oxidation state. It is probably in the MoO3 oxidation state. Al2p spectrum shows a single peak, due to Al3+ (Al2O3) oxidation state (fig. 3.9d). The asymmetric shape of the peak in the lower energy part of the spectrum may be due to charge effects [102].
Figure 3.10 The peak of oxygen determined with XPS.
Fig. 3.10 shows a rather large peak with FWHM = 2.6 (FWHM: Full Width at Haft Minimum). This is the most difficult to determined exactly the state of oxygen, since there are 2 peaks, one related to O2- from the different chemical species present in the mixture, the other may be due to OH- or H2O.
The quantities of chemical elements in composition have been calculated from the XPS spectrum. They are reported in table 3.3.
Table 3.3 Raw semi-quantification for T1 sample
Fe Mo Al O C
1% 1.2% 17.2% 46.4% 34.2%
The results of XPS experiment:
- Fe is clearly detected, though it is in low quantities.
- Oxidation states of Fe and Mo have been characterized (Fe2O3 and MoO3, respectively.