Cobalt Catalyst Characterization for Methane Decomposition and Carbon Nanotube Growth

Abstracts

Several analytical techniques, such as N₂ physisorption, X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), temperature programmed reduction (TPR) and chemisorption were employed to characterize the structure of Co/γAl₂O₃ catalysts used for the production of carbon nanotubes by methane vapor deposition. The catalysts were studied after the calcination step and presented three main cobalt species, CoAl₂O₄, CoO and Co₃O₄. The CoAl₂O₄ species were well dispersed and were reduced only at high temperatures, rendering them inactive for the carbon nanotube production. In the case of the 1%Co/Al₂O₃ catalyst, the main cobalt species found was CoAl₂O₄. However, increasing the cobalt content in the catalysts led to a higher formation of Co3O4 as compared to CoO and CoAl₂O₄ species. The 2 and 3%Co/Al₂O₃ catalysts showed particle agglomeration during the pretreatment step that decreased selectivity towards nanotube production. The 4%Co/Al₂O₃ catalyst did not show particle agglomeration and presented a higher selectivity to carbon nanotube production, 71%, mainly multi-walled carbon nanotubes (MWNT).

carbon nanotubes; cobalt; catalyst support; chemical vapor deposition; methane

Varias técnicas, como fisissorção, espectroscopia de fotoelétrons excitados por raios X (XPS), difração de raios X (XRD), redução à temperatura programada (TPR) e quimissorção foram usadas para a caracterização de catalisadores de Co/γAl₂O₃ para a produção de nanotubos de carbono por deposição de metano. Os catalisadores apresentaram três principais espécies de cobalto: CoAl2O4, CoO e Co₃O₄. As espécies de CoAl₂O₄ estavam bem dispersas nos catalisadores e apresentaram elevada temperatura de redução, fazendo com que estas espécies sejam inativas para a produção de nanotubos. Para o catalisador 1%Co/Al₂O₃, a principal espécie encontrada foi o CoAl₂O₄. Contudo, com o aumento do teor de cobalto nos catalisadores, ocorreu um aumento na formação das espécies Co₃O₄ em relação às espécies CoO e CoAl₂O₄. Os catalisadores 2 e 3%Co/Al₂O₃ apresentaram aglomeração das partículas de cobalto após a etapa de redução e baixa seletividade para produção de nanotubos. Entretanto, o catalisador 4%Co/Al₂O₃ não apresentou aglomeração, resultando em uma melhor seletividade para formação de nanotubos, 71%, principalmente do tipo nanotubos de carbono de paredes múltiplas (MWNT).

Introduction

Carbon nanotubes (CNT) exhibit unique physical and chemical properties that have opened a vast number of applications and new ones are still under development. However, for the incorporation of CNT into everyday materials, it is important to control their growth. The production of CNT with pre-determined specifications is an essential requirement for their industrial use. The investigation into CNT production by catalytic chemical vapor deposition (CVD) has attracted great attention and has become the most popular synthesis method due to its easy setup and scale-up for mass production.^1,2

Many researchers^3-5 have studied different catalyst formulations and operating conditions for CVD process improvement. Cobalt-supported catalysts have shown high selectivity towards CNT production. The nature and morphology of the cobalt species affects the catalyst activity and selectivity on CNT production. So, it is important to understand the influence of these cobalt species on carbon nanotube production in order to develop a strategy to maximize the catalyst selectivity and activity. In our previews work,⁶ among various catalyst formulations investigated, cobalt catalysts supported on alumina exhibited the highest carbon nanotube yield. Therefore, this contribution has focused on the characterization of the cobalt species present in a cobalt/alumina system for each step of CNT growth and explored their influence in CNT synthesis.

In our CNT production method, there are two main steps: reduction of the oxide catalyst followed by methane decomposition on the reduced catalyst for carbon nanotube growth. Therefore, 1, 2, 3 and 4 wt. % Co/Al₂O₃ catalysts were studied in three different states, as an oxide, reduced and passivated.

The catalysts were investigated using energy-dispersive X-ray spectroscopy (EDX), X-ray diffraction (XRD), N₂ physisorption, temperature programmed reduction (TPR), CO and H₂ chemisorption and X-ray photoelectron spectroscopy (XPS). The combination of these techniques can detail the structure of the catalysts, which can be used to clarify the activity and selectivity toward carbon nanotube formation.

CNT characterization was performed by Raman spectroscopy, temperature programmed oxidation (TPO), scanning electron microscopy (SEM) and transmission electron microscopy (TEM).

Experimental

Catalyst preparation

The Al₂O₃ support was obtained by drying bohemite (Catapal-A, Sasol, degree of purity: 99.9%) in air at 120 ºC for 16 h and then calcining in air at a heating rate of 10 ºC min^-1 up to 550 ºC for 2 h. A series of 1, 2, 3 and 4 wt. % Co/Al₂O₃ catalysts was prepared by incipient wetness impregnation using a solution of Co(NO₃)₂.6H₂O (Merck, degree of purity: 99.8%). After impregnation, the catalysts were dried in air at 120 ºC for 16 h and then calcined in air for 2 h at 550 ºC.

Catalyst characterization

The oxide catalysts were studied after the calcination step. For the reduced state, the oxide catalysts were reduced under H₂ flow from room temperature to 700 ºC at a heating rate of 10 ºC min^-1 and were then analyzed. For the passivated state, the samples were reduced as described above and then the catalysts were cooled to room temperature in He flow. Finally, they were passivated with 1.0% O₂/He (30 mL min^-1) during 2 h at room temperature.

Measurements of surface area, pore volume and pore diameter distribution were performed in a Micromeritics ASAP 2010 apparatus by N₂ adsorption. The oxide catalysts were evacuated at 220 ºC during 4 h before analysis. Pore volume and pore diameter distribution were calculated from the N₂ desorption curve using the Barrett-Joyner-Halenda (BJH) method.

The chemical compositions of the oxide catalysts were determined by EDX in vacuum using an EDX-720 Shimadzu apparatus.

XPS experiments of the oxide and passivated catalysts were performed with a Thermo Scientific ESCALAB 250Xi with monochromatic Al Kα X-rays with a spot size of 650 µm. The base pressure inside the analysis chamber was 1 × 10^-9 mbar or lower. The samples were mounted on a sample holder by means of a double-sided adhesive carbon tape. For the survey spectra an energy step size of 1.0 eV and a pass energy of 100 eV were used. For the individual element spectra, Co2p, O1s and Al2p, an energy step size of 0.05 eV and a pass energy of 25.0 eV were used. The flood gun was used to neutralize charge buildup on the surface of the samples. The spectra were analyzed and peak fitted with a linear type background and a Gaussian product function.

XRD experiments of the oxide and passivated catalysts were performed in a Rigaku Miniflex diffractrometer with monochromatic Cu Kα (1.540 Å) radiation with a scan rate of 0.05º min^-1 in the range of 2θ = 2.00 to 80.00º. The average cobalt particle sizes were calculated from the most intense Co₃O₄ line, 2θ = 36.9º, using the Scherrer formula.

Raman spectroscopy was used for the characterization of the oxide catalysts and carbon deposits on the catalysts after the reaction step. Raman spectra were taken in a Confocal Raman Microscope alpha 300, Witec, using a 50× objective lens and green laser with 532 nm wavelength. The integration time was 1.0 s and the number of scans was 500.

TPR experiments were performed in a U-shaped tubular quartz reactor coupled to a quadrupole mass spectrometer (Omnistar, Balzers). The oxide catalysts, approximately 0.5 g, were dried in He for 30 min at 150 ºC before TPR. After cooling to room temperature, a mixture of 5% H₂/Ar flowed (30 mL min^-1) through the catalyst and the temperature was raised at a heating rate of 10 ºC min^-1 up to 1000 ºC.

H₂ and CO chemisorptions were performed using a volumetric method in a Micromeritcs ASAP 2010 device. The pretreatment of the oxide catalysts consisted of drying at 220 ºC for 30 min under a 30 mL min^-1 He flow before the reduction. After cooling to room temperature, the reduction step was performed. The temperature was raised at a heating rate of 10 ºC min^-1 up to 700 ºC under 30 mL min^-1 of H₂. Then, the reduced catalysts were outgassed under vacuum at 700 ºC before cooling to 150 ºC, where the H₂ chemisorption measurements were performed. Then, the catalysts were evacuated at 500 ºC, and finally the CO chemisorption measurements were done at 35 ºC. H₂ and CO chemisorption on cobalt are performed at temperatures higher than room temperature as they are activated.⁷ Total and reversible isotherms were measured at the pressure range of 10-300 mmHg. The amount of H₂ and CO adsorbed in each measurement was determined by extrapolating the linear part of the isotherm to zero pressure. It is assumed that the adsorption stoichiometries for H₂ and CO are H/Co = 1 and CO/Co = 1, respectively. The Co particle size was calculated supposing spherical Co particles. For H₂ chemisorption, these calculations were performed assuming total chemisorption as recommended by Reuel and Bartholomew.⁸

CNT production

For the CNT production by methane CVD, 1.0 g of catalyst was inserted into a quartz boat and then placed in a horizontal tubular quartz reactor. The catalyst was heated under 200 mL min^-1 H₂ flow from room temperature to 700 ºC at a heating rate of 10 ºC min^-1. Then, the H₂ flow was switched to 200 mL min^-1 methane for 30 min. Finally, the sample was cooled to room temperature under Ar flow.

CNT characterization

TPO experiments were performed in the same apparatus described for the TPR experiments. The samples (0.2 g) were dried in He for 30 min at 200 ºC before TPO. Afterwards, the sample was cooled to room temperature, a 5% O₂/He mixture flowed through the sample at 30 mL min^-1 as the temperature was raised at a heating rate of 10 ºC min^-1 up to 1000 ºC. There was always O₂ in excess to avoid the formation of carbon monoxide. The mass of carbon present in each sample was determined by the evolution of CO₂.

CNT morphology was analyzed by SEM images. SEM images were performed in a Zeiss EVO MA and a Jeol JSM-5800 LV scanning electron microscopes with electron beam energy in the 10-30 kV range.

The TEM images were obtained in a FEI Tecnai F30 transmission electron microscope. For this experiment, the samples were dispersed in isopropanol and ultrasonicated for 15 min. Then, a few drops of the resulting suspension were deposited on a grid and subsequently evaporated.

Results and Discussion

N₂ physisorption

N₂ physisorption measurements were performed to determine the changes in surface area and pore diameter distribution between the support and catalyst after the calcination step. Table 1 shows the textural properties of the support and cobalt catalysts. The surface area and the average pore diameter of the alumina are typical of γ-phase, as found by Rane et al..⁹

Table 1
Textural properties and chemical composition of the catalysts

Sample	Surface area / (m2 g-1)	Pore volume / (cm3 g-1)	Average pore diameter / nm	Bulk Coa / wt.%	Surface Cob / wt.%	Fraction of Co as CoAl2O4b
Al2O3	282	0.64	7.1	-	-	-
1Co/Al2O3	266	0.63	7.4	0.92	1.10	0.93
2Co/Al2O3	257	0.61	7.5	2.02	0.54	0.82
3Co/Al2O3	256	0.61	7.5	2.99	0.56	0.69
4Co/Al2O3	248	0.59	7.8	3.96	0.74	0.52

• ^a
  Measured by EDX;
• ^b
  measured by XPS.

Higher cobalt loading catalyst showed lower surface area and pore volume. This may be attributed to the blocking of narrow pores by the cobalt oxide particles. Since the narrow pores are blocked, there is an increase in average pore diameter as shown in Figure 1 and Table 1.

Figure 1
Support and catalyst pore size distribution.

Energy-dispersive X-ray spectroscopy

The catalyst chemical composition was analyzed by EDX. The cobalt fraction is consistent with the nominal value used in the catalyst preparation. Table 1 shows the cobalt composition of the catalysts.

X-ray photoelectron spectroscopy

XPS can be used to determine the surface chemical composition of the catalyst. Two survey spectra were obtained at different points for each catalyst and they showed the same profile. This indicates a uniform cobalt distribution on the support. The survey spectrum detected four elements: cobalt, aluminum, oxygen and carbon. The carbon comes from adventitious hydrocarbons nearly always present on samples, introduced from the laboratory environment or from adhesive carbon tape.

The cobalt composition found by XPS is a function of its depth and surface distribution. So, when cobalt particles are concentrated in the top layers of the catalyst or they are highly dispersed on the support, the intensity of the cobalt X-ray photoelectron is increased. 1Co/Al₂O₃ showed approximately the same surface cobalt concentration as its bulk value, Table 1. This is an indicative that the cobalt particles are uniformly distributed within the catalyst pellet.

Usually, three different cobalt species are present in a Co/Al₂O₃ catalyst: Co₃O₄, CoO and CoAl₂O₄. Each of these cobalt species shows a different color; cobalt aluminate is light blue, CoO is olive greenish-brown and Co₃O₄ is black.^10,11 1Co/Al₂O₃ catalysts presented a light blue color, indicating that most of the cobalt is in the form of cobalt aluminate. Increasing the cobalt loading of the catalysts, they became darker until turning black for 4%Co/Al₂O₃. So, the amount of Co₃O₄ species increased with cobalt loading and XPS helped in identifying these species.

The identification of cobalt chemical states depends on the accurate determination of the binding energy lines. The flood gun technique was used to neutralize charge build-up on the sample surface. However, the flood gun may provide more electrons to the catalyst than those leaving it. As a result, the whole catalyst surface may charge up due to the energy of the flood gun beam. For this reason, the peak energies were corrected by reference to the C1s line at 284.8 eV. In our experiments, there was an excessive charge compensation making the peaks shift to lower binding energy at values of 0.5, 0.4, 0.3 and 0.2 eV for the catalysts with cobalt loadings of 1, 2, 3 and 4 wt. %, respectively. Other authors, such as Hilmen et al.,¹² have corrected the peak energy by reference to Al2p. In our results, there was no significant difference between the correction by reference to the C1s line or to the Al2p line.

The binding energy of the cobalt species present on the catalysts can be compared to the binding energy of reference compounds, such as Co₃O₄, CoO and CoAl₂O₄.^3,12,13 According to Hilmen et al.,¹² the binding energy of the Co2p_3/2 peak is found at 780.2 eV for a Co₃O₄ bulk sample and its asymmetric shape can be explained due the slightly higher binding energy of Co²⁺ in relation to Co³⁺ (780.3 and 779.5 eV, respectively). For a bulk CoO sample, Herreraet al.³ found a binding energy of 780.5 eV for the Co2p_3/2 peak. As these cobalt species show approximately the same binding energy, it is hard to distinguish these peaks precisely. However, the Co2p_3/2 peak of CoAl₂O₄ is found at higher binding energy, 782.0 eV, than the binding energy of the 2p_3/2 peak of Co₃O₄ and CoO. In this way, it is possible to distinguish the cobalt aluminate Co2p_3/2 peak from the other cobalt species. Therefore, the amount of CoAl₂O₄ could be estimated by deconvolution of the Co2p_3/2 peak into two mixed Gaussian-Lorentzian curves centered at a fixed binding energy peak, 780 eV for CoO and Co₃O₄ species and 782.0 eV for CoAl₂O₄. Figure 2 shows the XPS spectra of the oxide catalyst for the Co2p region and Table 1 shows the fraction of cobalt aluminate present on the catalysts.

Figure 2
XPS spectra of the oxide catalyst in the Co2p region. (a) 1Co/Al₂O₃; (b) 2Co/Al₂O₃; (c) 3Co/Al₂O₃; (d) 4Co/Al₂O₃. The vertical dot lines are binding energy reference compounds: Co₃O₄ (780.2 eV) and CoAl₂O₄ (782.0 eV).¹²

The Co2p_3/2 peak of the 1Co/Al₂O₃ catalyst has the same binding energy of the CoAl₂O₄ bulk sample, as found by Hilmen et al..¹² It shows that most of the cobalt species on the 1Co/Al₂O₃ catalyst is CoAl₂O₄. Increasing the cobalt content on the catalysts, the binding energy of the Co2p_3/2 peak shifted to lower energy and the satellite peak intensity decreased. The first effect shows an increase of the CoO and Co₃O₄ fraction in relation to CoAl₂O₄ and the second effect points to an increase of Co₃O₄ species in relation to CoO. The satellite peak intensity is weaker for Co₃O₄ than for CoO and centered about 10 eV from the main peaks.¹³ In our results, the satellite peak could only be observed in the case of the 1Co/Al₂O₃ catalyst, Figure 2.

The Co₃O₄ and CoAl₂O₄ species represent the two limit cases of cobalt/alumina interaction in a catalyst. The Co₃O₄ species has no any interaction with alumina while CoAl₂O₄ shows complete spinel formation.¹²

Kerkhof and Moulijn¹⁴ have shown that the supported phase/support signal intensity ratio is related to the dispersion of the supported phase. In this model, the catalyst is supposed to consist of several sheets of the support with cubic particles of equal size in between. Thus, the oxide cobalt particle size can be estimated from the Co2p_3/2 and Al2p intensity ratio, (I_Co/I_Al)_crystallite, according to equation 1:

where, α = , is the cube edge length of the average cobalt particle and λ_Co is the mean escape depth of the photoelectrons from cobalt passing through cobalt oxide.

The predicted monolayer intensity of Co and Al were calculated using the Kerkhof and Moulijn model,¹⁴ according to equation 2:

where β₁ = t / λ_(Al/Al) and β₂ = t / λ_(Co/Al). λ_(Al/Al) is the mean escape depth of the photoelectrons from aluminum passing through the alumina, λ_CoAl is the mean escape depth of the photoelectrons from cobalt passing through the alumina and t is mean thickness of the alumina. The term (n_Co/n_Al)_bulk is the bulk atomic ratio of cobalt and aluminum from the catalysts. The mean escape depth of the photoelectrons used were obtained from Penn.¹⁵

The detector efficiency D is a function of the kinetic energy of the electrons and σ is the photoelectron cross sections. These parameters were substituted by the atomic sensitivity factor from the Avantage Software library, 12.62 for Co2p_3/2 and 0.537 for Al2p.

For the 1Co/Al₂O₃ catalyst, the ratio between the measured Co2p_3/2 and Al2p intensity ratios and the predicted monolayer intensity ratio is 1.28 as shown in Table 2. As this ratio cannot be higher than 1, this result indicates that cobalt aluminate is not dispersed as in the Kerkhof and Moulijn model. For the other catalysts, the average CoAl₂O₄ particle size was estimated. The cobalt aluminate particle size increased with the cobalt content. As the cobalt Co2p_3/2 peak from CoO and Co₃O₄ have approximately the same binding energy, it was not possible to separate these peaks precisely. Hence, the average particle size of CoO and Co₃O₄ were estimated together. The average cobalt oxide size also increased with cobalt content and they are larger than the CoAl₂O₄ particles. The cobalt particle size is shown in Table 2.

Table 2
Average cobalt particle size estimated by XPS intensities, Co₃O₄ crystallite average size calculated by XRD

Sample	(ICo/IAl)cryst. / (ICo/IAl)mono	CoAl2O4 particle sizea / nm	Cobalt oxide particle sizea / nm	Co3O4 crystallite sizeb / nm	Fraction of metallic phasea
1Co/Al2O3	1.28	-	-	-	0
2Co/Al2O3	0.29	2.4	2.4	-	1.5
3Co/Al2O3	0.20	3.0	4.1	10.0	2.4
4Co/Al2O3	0.20	3.2	4.3	14.3	8.3

• ^a
  Measured by XPS;
• ^b
  measured by XRD.

For the 1CoAl₂O₃ catalyst, a highly dispersed cobalt aluminate may be formed by cobalt atom migration into the bulk alumina matrix, as found by Liotta et al.¹⁶ and Jongsomjit et al..¹⁷

The 1Co/Al₂O₃ catalyst presented only CoAl₂O₄ species as found by XPS, a surface analysis, and by TPR, a bulk analysis (see TPR section, below). This catalyst showed approximately the same CoAl₂O₄ concentration on the surface (XPS measurements) and in the bulk phase, as found by EDX analysis (Table 1). Application of the Kerkhof and Moulijn model has also shown that the cobalt aluminate species remain highly dispersed in the other catalysts as in the case of 1Co/Al₂O₃. In this way, the CoAl₂O₄ concentration on the surface was assumed as equal to the bulk concentration for all studied catalysts and the total amount of cobalt as CoAl₂O₄ present in the catalysts was determined by multiplying the fraction of Co as CoAl₂O₄ estimated by XPS by the weight fraction of bulk Co measured by EDX.

Increasing the cobalt content in the catalysts, there was an increase in the total amount of CoAl₂O₄ up to 2.1 wt. % on the 3Co/Al₂O₃ catalyst. This shows that there is a maximum limit between 2 and 3 wt. % Co for the CoAl₂O₄ formation on the cobalt/alumina catalyst, when it is prepared as in the synthesis conditions described in the experimental section.

X-Ray diffraction

XRD analysis was used to determine the crystalline phases and to estimate the cobalt oxide crystallite size. Figure 3 shows the XRD patterns of the support and oxide catalysts. The diffraction peak at 40.0 and 66.7º are those of γ-Al₂O₃ and they can be observed in all samples.

Figure 3
XRD patterns of oxide catalysts: (a) Al₂O₃; (b) 1Co/Al₂O₃; (c) 2Co/Al₂O₃; (d) 3Co/Al₂O₃; and (e) 4Co/Al₂O₃. □ Co₃O₄ phase; ● γ-Al₂O₃ phase.

The 3 and 4Co/Al₂O₃ catalysts showed diffraction peaks at 31.3, 36.9 and 59.5º. These diffraction lines can be related to Co₃O₄ or CoAl₂O₄ species.¹⁷ In this way, it was not possible to distinguish these two phases by XRD. However, CoAl₂O₄ species are well dispersed and there is a maximum on its formation (2.1 wt. %) in the 3Co/Al₂O₃ catalysts, as found by XPS. This way, the increase in the cobalt diffraction peak intensity from 3 to 4Co/Al₂O₃ catalysts on the XRD patterns may be attributed to the Co₃O₄ species. The diffraction peaks of CoO, 42.6 and 61.8º,¹⁷ could not be observed in the XRD patterns. It may due to its low loading, which can lead to a very small particle size or the CoO particles were amorphous.

The average Co₃O₄ crystallite size was calculated from the most intense Co₃O₄ peaks at 36.9º using the Scherrer equation. These results can be observed in Table 2.

The average crystallite sizes found by XRD were quite larger than the particle size found by XPS. In the case of XPS, the particle size was estimated from the Co2p_3/2 peak associated with photoelectrons leaving the CoO and Co₃O₄ particles, not only from the crystalline Co₃O₄ particles. Thus, the particle size calculated by XPS is probably an average of small CoO and large Co₃O₄ particles.

Temperature programmed reduction

TPR can help with the identification and quantification of the cobalt oxide species supported on alumina. The TPR profile for 1Co/Al₂O₃ didn't show any significant hydrogen consumption, Figure 4a. This shows that there was only irreducible cobalt species present on this catalyst, in the form of CoAl₂O₄, as suggested by Jongsomjit et al..¹⁷ XPS analysis showed that 93 wt. % of these species are cobalt aluminate. The presence of this species on the catalysts is not desired, since only metallic cobalt particles are able to decompose methane and grow carbon nanotubes.

Figure 4
TPR profiles of catalysts (a) 1Co/Al₂O₃; (b) 2Co/Al₂O₃; (c) 3Co/Al₂O₃; and (d) 4Co/Al₂O₃. The reduction was conducted under 5% H₂/Ar, using a heating rate of 10 oC min^-1.

The TPR profile of the 2Co/Al₂O₃ catalyst showed four small reductions peaks, suggesting that there are four different reducible cobalt species. The peak at 364 ºC is attributed to the reduction of an amorphous cobalt oxide similar to bulk Co₃O₄.¹⁸ The peak around 507 ºC is attributed to the two-step reduction of Co₃O₄. This reduction temperature is higher than the reduction temperature of a bulk Co₃O₄ sample. This shift to higher temperatures demonstrates that there is an interaction between the support and the cobalt oxide particles.¹⁹ The small peak at 737 ºC is associated with the reduction of CoO linked to alumina¹⁹ and the reduction peak at 1000 ºC is attributed to reduction of cobalt aluminate.²⁰

The irreducible cobalt aluminate present in the 1Co/Al₂O₃ catalyst may have different structure from the reducible aluminate present in the 2Co/Al₂O₃ catalyst. The cobalt aluminate spinel may be a partly inverse spinel, in which some Co²⁺ ions are placed in unstable octahedral positions, or a normal spinel, with Co²⁺ ions placed in stable tetrahedral positions. Because of their different stability, Co²⁺ ions in partly inverse spinel are reduced at lower temperatures than in normal spinel.

The reduction profile for the 3Co/Al₂O₃ catalyst also showed four different reducible cobalt species, as in 2Co/Al₂O₃. However, the fraction of each reduced cobalt species was different. The fraction of Co₃O₄ (peak at 507 ºC) was higher for 3Co/Al₂O₃ than for 2Co/Al₂O₃. This peak was also intense in the TPR profile of the 4Co/Al₂O₃ catalyst, in agreement with its X-ray diffractogram.

The degree of reduction is an important parameter on catalyst activity and different cobalt oxide species require different amounts of hydrogen to be reduced to its metallic form. The fraction of the cobalt present as Co₃O₄ species can be estimated from the hydrogen consumption from 130 to 645 ºC. This reduction corresponds to the following reaction: Co₃O₄ + 4H₂ → 3Co + 4H₂O. Thus, a molar ratio of 1.33 H₂/Co is required. The CoO and CoAl₂O₄ species demand a molar ratio of 1H₂/Co for their reduction.^18,19 In this way, it was possible to estimate the total hydrogen consumption for a complete catalyst reduction and the degree of reduction was then calculated. Table 3 shows the degree of reduction for the several catalysts.

Table 3
Results from TPR. Oxide catalyst degree of reduction, distribution of cobalt species and degree of reduction of the reduced catalysts (after pretreatment reduction step)

Sample	Degree of reduction / %	Fraction of Co3O4	Fraction of CoO and CoAl2O4	Pretreatment degree of reduction / %
1Co/Al2O3	0	0	1	0
2Co/Al2O3	27	0.07	0.93	17
3Co/Al2O3	30	0.16	0.84	20
4Co/Al2O3	31	0.24	0.76	22

The degree of reduction increased with the cobalt content in the catalyst. However, there was a large cobalt fraction that was not reduced in all studied catalysts. The strong interaction between cobalt and alumina can form surface and bulk spinels that are not easily reduced.²¹ The trend of increasing degree of reduction with the cobalt content was previously observed for the Co/Al₂O₃ catalysts.²¹

While all Co₃O₄ was reduced to its metallic form, only a small fraction of CoO and CoAl₂O₄ was reduced from 645 to 1000 ºC. In this way, the overall fraction of CoO and CoAl₂O₄ could not be determined by TPR. However, the fraction of CoO can be calculated by the subtraction of the Co₃O₄ fraction determined by TPR from the CoAl₂O₄ fraction estimated by XPS. XPS was used to estimate the CoAl₂O₄ concentration, considering that this species is uniformly distributed through the catalyst, as discussed before. Figure 5 shows the distribution of cobalt species for the catalysts.

Figure 5
Cobalt species distribution for the catalysts. □ Species that were reduced; ◊ species that were not reducible during TPR experiments.

The increase of cobalt loading on the catalysts led to a higher formation of reducible Co₃O₄ as compared to reducible CoO and CoAl₂O₄ species. From Figure 5 we can also note that the weight percentage of cobalt as irreducible CoAl₂O₄ did not increase much, increasing with the cobalt content up to the 3Co/Al₂O₃ catalyst while the formation of irreducible CoO keeps increasing. These changes in composition affect the catalyst color and they became darker with the increase of Co₃O₄ species.

Chemisorption

A better representation of the cobalt particle as it stands before the carbon nanotube reaction can be done by its characterization in the reduced or in the passivated state. The chemisorption measurements may be used to estimate the average metallic particle size. For this, it is necessary to know the catalyst degree of reduction at the beginning of reaction. So, the pretreatment degree of reduction was determined by a new TPR after the reduction step. The difference between the hydrogen consumption for reduction of the oxide catalysts and for the partially reduced catalysts is related to the hydrogen consumption during the reduction pretreatment step. Table 3 shows values of the pretreatment degree of reduction of the catalysts. The pretreatment degree of reduction increased with cobalt loadings up to 22%. After pretreatment, the catalysts show values of degree of reduction similar to the degree of reduction observed after the standard TPR. So, an increase in the temperature of catalyst reduction during the pretreatment reduction step would not increase the final degree of reduction. Additionally, static H₂ and CO chemisorption measurements were used to determine the dispersion and average particle size of the metallic cobalt particles. These reduced species are related to the catalyst activity and structure of carbon nanotube growth. There is a correlation between the size of the catalyst particles and the CNT diameter.^22-24 The average particle size was calculated from total H₂ uptake and from irreversible CO uptake using a stoichiometric adsorption ratio of 2 and 1, respectively. Table 4 shows the total and irreversible H₂ and CO uptakes and average particle size.

Table 4
Total and irreversible H₂ and CO chemisorption, dispersion and particle sizes of catalysts obtained from chemisorption

Sample	H2 / (μmol g-1 cat)			CO / (μmol g-1 cat)		Total H/Co	Irrever. CO/Co	Particle size / nm
	Total	Irrever.		Total	Irrever.			Total H2	Irrever. CO
1Co/Al2O3	0.0	0.0		0.7	0.0	-	-	-	-
2Co/Al2O3	2.6	1.1		13.7	10.5	0.010	0.031	12.2	5.5
3Co/Al2O3	4.5	1.9		25.8	20.7	0.018	0.041	11.2	4.9
4Co/Al2O3	7.2	3.1		44.6	30.6	0.021	0.046	9.8	4.6

H₂ and CO uptakes increased with cobalt loading and higher values were obtained for CO uptake, as previously reported by Reuel and Bartholomew.⁸ Besides, the total H/Co and irreversible CO/Co ratios have shown some increase with increasing metal loading. In this way, the 4Co/Al₂O₃ catalyst showed better metallic dispersion and smaller metallic particle size than 2 and 3 wt. % cobalt catalysts. However, in the oxidized state, the particle size increased with the increase of cobalt loading. Nevertheless, the 2 and 3 wt. % cobalt catalysts showed cobalt oxide particles that are reduced at low temperature (see TPR profile). These metallic particles may have agglomerated on the reduction pretreatment step.

To clarify the cobalt agglomeration during the reduction pretreatment, the reduced catalysts were re-oxidized with 1.0% O₂/He (30 mL min^-1) during 2 h at 550 ºC. Then, these catalysts were analyzed by XRD. The particle size of the 1 and 2 wt. % cobalt catalysts could not be estimated due the low Co₃O₄ peak signal. However, the particle size of 3Co/Al₂O₃ increased from 10.0 to 14.3 nm and the 4Co/Al₂O₃ catalyst didn't change much its particle size, 14.2 nm. These results support the particle agglomeration suggestion during the reduction pretreatment.

These re-oxidized samples were also studied by XPS measurements, Figure 6. There was no significant difference between the XPS spectra of 1Co/Al₂O₃ in the oxide and passivated state. This result agrees with TPR findings, where no reduction was observed for the 1Co/Al₂O₃ oxide catalyst. Thus, there was no chemical change in its composition after the reduction and passivation steps. The XP spectrum of the other passivated catalysts showed small differences in their profile in relation to oxide profile. These differences can be observed in the inset graph of Figure 6. The 2, 3 and 4Co/Al₂O₃ catalysts showed two main peaks in the inset. The first peak, at 778.2 eV, represents the increase of the metallic phase,^3,12 and the second peak (negative), at 780.3 eV, represents the decrease of cobalt as oxide species in the passivated catalyst in relation to the oxide catalysts.

Figure 6
XPS spectra of the passivated catalysts in the Co2p region. The oxide catalysts were reduced in H₂ at 700 oC and then exposed to 1% O₂/He for 2 h at room temperature. (a) 1Co/Al₂O₃; (b) 2Co/Al₂O₃; (c) 3Co/Al₂O₃; and (d) 4Co/Al₂O₃. The vertical dot lines are binding energy reference compounds: Co⁰ (778.2 eV), Co₃O₄ (780.2 eV) and CoAl₂O₄ (782.0 eV).12 Inset: difference on the XPS profile of the oxide and passivated catalysts.

The fraction of metallic phase, R, can also be obtained by using the following equation:²⁵

The fraction of metallic phase increased with the cobalt loading, Table 2. However, the degree of reduction found by XPS is significantly lower than that calculated by TPR. The passivation step produces a cobalt oxide shell on the metallic cobalt particle and the cobalt fraction on the oxide shell is higher on small particles. Moreover, XPS is a surface method while TPR is a bulk method.

Carbon nanotube production and characterization

TEM

Carbon nanotube production was characterized by SEM and TEM. The SEM images showed low-density carbon deposits in the case of the 1, 2 and 3Co/Al₂O₃ catalysts. However, the 4Co/Al₂O₃ catalyst showed a high carbon density, Figure 7a. Increasing the SEM magnification, the image shows that most of the surface is covered by carbon filaments, Figure 7b. From Figure 7c along with other images, the filament average diameter was estimated as 13 nm (73 filaments measured). The average diameter of metallic cobalt particles found by H₂ chemisorption was approximately the same of the average diameter of the filaments, 14.2 nm. TEM images show that theses filaments are carbon nanotubes and most of them are multi-walled carbon nanotubes (MWNT). Figure 7d shows MWNT growth on the 4Co/Al₂O₃ catalyst.

Figure 7
Images of carbon nanotubes prepared on 4Co/Al₂O₃ catalysts. (a), (b) and (c) SEM images at different magnifications; (d) TEM image.

TPO

The total carbon yield during the reaction step was defined as the mass of carbon deposited per mass of catalyst. 1Co/Al₂O₃ didn't show any significant carbon deposition. In this catalyst, most of the cobalt is present as cobalt aluminate and it could not be reduced during the reduction pretreatment. Table 5 shows the total carbon yield for the several catalysts. The carbon yield is proportional to the metallic surface area. The 4Co/Al₂O₃ catalyst showed the highest production, 37.5%.

Table 5
TPO and Raman spectroscopy data of carbon deposits on the catalysts

Carbon deposit on catalyst	Yield / %	Selectivity / %			D/G ratio
		Amorphous	SWNT	MWNT
1Co/Al2O3	0	-	-	-
2Co/Al2O3	12.2	85	11	4	1.3
3Co/Al2O3	15.1	80	11	10	1.0
4Co/Al2O3	34.1	29	29	42	0.7

Combining the data obtained from TPO, Raman and electron microscopy, it was possible to estimate the amount of each carbon species produced in the reaction step. Three main different species can be produced in the carbon nanotube reaction step. These different carbon species oxidize at different temperatures. The TPO profiles of the carbon species present on the catalysts are illustrated in Figure 8. The 2 and 3Co/Al₂O₃ catalysts showed low selectivity toward carbon nanotubes. In their profiles, there is a large peak at 380 ºC related to amorphous carbon oxidation. These catalysts have shown metallic particle agglomeration in the reduction pretreatment. These agglomerated particles may not have the adequate shape or size for carbon nanotube growth. Table 5 shows the catalyst activity and selectivity of catalysts.

Figure 8
TPO profiles of the carbon species present on (a) 2Co/Al₂O₃; (b) 3Co/Al₂O₃; and (c) 4Co/Al₂O₃. (d) TPO profiles of the carbon species present in 4Co/Al₂O₃ after CNT reaction and oxidation step under 1% O2/He (30 mL min^-1) at a heating rate of 2 oC min^-1 up to 350 ºC.

Raman

Raman spectroscopy provides information about the structure of carbon nanotubes and other carbon species. The tangential mode, called G band, at 1590 cm^-1 is used to identify the presence of ordered carbon-like graphite and carbon nanotubes. The D band at 1350 cm^-1 is related to defects on the carbon nanotube structure and the presence of amorphous carbon.^26-28 The intensity of the D band relative to the G band has been used as qualitative measurement of the formation of undesirable forms of carbon on nanotube production. Figure 9 shows the Raman spectra of carbon deposits formed on the catalysts.

Figure 9
Raman spectra of the carbon deposits on (a) 2Co/Al₂O₃; (b) 3Co/Al₂O₃ and (c) 4Co/Al₂O₃. (d) Spectrum of the carbon deposits on 4Co/Al₂O₃ after CNT reaction and oxidation step under 1% O₂/He (30 mL min^-1) at a heating rate of 2 ºC min^-1 up to 350 ºC. (e) Spectrum of the 4Co/Al₂O₃ catalyst. RBM: radial breathing mode.

The carbon deposits on the 2 and 3 Co/Al₂O₃ catalysts showed high D/G intensity ratios. These results indicate the presence of amorphous carbon and/or poorly structured carbon nanotubes, which were not observed on the SEM images. The TPO profile of these samples showed that most of the carbon deposits were amorphous carbon.

The D/G intensity ratio of carbon deposits on 4Co/Al₂O₃ was lower than on the 2 and 3Co/Al₂O₃ catalysts. The TPO profile also showed a decrease in the amorphous carbon fraction.

The influence of amorphous carbon in the Raman spectrum was verified by its oxidation for the 4Co/Al₂O₃ catalyst. The oxidation consisted of a flow of 1% O₂/He (30 mL min^-1) at a heating rate of 2 ºC min^-1 up to 350 ºC. Then, the sample was cooled under He flow until room temperature and a new TPO profile and a Raman spectrum were acquired, Figure 8d and Figure 9d, respectively. The new TPO profile showed a smaller fraction of amorphous carbon and the remnant carbon species deposited on the catalysts did not change their oxidation profile significantly. In the case of the Raman experiments, the D/G intensity improved to 0.5, indicating the amorphous carbon was removed by the oxidation at 350 ºC.

The carbon deposits on the 4Co/Al₂O₃ catalyst showed two small peaks at 167 and 183 cm^-1. These peaks are associated with single-walled carbon nanotubes (SWNT) and they are called radial breathing mode (RBM) peaks. TPO showed that 29% of carbon deposits on 4 Co/Al₂O₃ were SWNT. Some oxide species, like CoO, Co₃O₄ and CoAl₂O₄ can also show Raman peaks in the RBM region. In order to identify the cobalt oxide peaks, a Raman spectrum of the 4Co/Al₂O₃ oxide catalyst was acquired, Figure 9e. There was not any overlap between the RBM peaks and the cobalt oxide peaks. So, the peaks found at 167 and 183 cm^-1 on spectra c and d in Figure 9 are from SWNT.

Conclusions

There are three main cobalt species in Co/γ-Al₂O₃ catalysts, CoAl₂O₄, CoO and Co₃O₄. Most cobalt in the 1Co/Al₂O₃ catalyst was present as CoAl₂O₄. This species was irreducible and highly dispersed on the support. Increasing the cobalt content of the catalysts led to a higher formation of Co₃O₄ as compared to the reducible CoO and CoAl₂O₄ species.

Catalysts with 2 and 3 wt. % of cobalt have cobalt oxide particles that reduced at low temperatures, lower than 400 ºC. These particles agglomerated during the pretreatment step. These particles may not have the shape or size for carbon nanotube production.

The 4Co/Al₂O₃ catalyst showed high selectivity toward carbon nanotube production, mainly MWNT. The average metallic particle size is approximately the size of the carbon nanotube diameter, with the same correlation found in the literature.

Acknowledgments

We would like to thank Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ Proc. E-26/110.128/2013) for financial support and also Instituto Militar de Engenharia (IME) for the SEM images. We dedicate this manuscript to the memory of Prof Roberto Fernando de Souza, who acted for the development of the catalytic community in Brazil.

References

• ¹
  Cheng, H. M.; Li, F.; Su, G.; Pan, H. Y.; He, L. L.; Sun, X.; Dresselhaus, M. S.; Appl. Phys. Lett.1998, 72, 3282.
• ²
  Zhou, W. Y.; Bai, X. D.; Wang, E. G.; Xie, S. S.; Adv. Mater.2009, 21, 4565.
• ³
  Herrera, J. E.; Balzano, L.; Borgna, A.; Alvarez, W. E.; Resasco, D. E.; J. Catal 2001, 204, 129.
• ⁴
  Irurzun, V. M.; Tan, Y. Q.; Resasco, D. E.; Chem. Mater 2009, 21, 2238.
• ⁵
  Wen, Q.; Zhang, R. F.; Qian, W. Z.; Wang, Y. R.; Tan, P. H.; Nie, J. Q.; Wei, F.; Chem. Mater 2010, 22, 1294.
• ⁶
  Oliveira, H. A.; Franceschini, D. F.; Passos, F. B.; J. Braz. Chem. Soc 2012, 23, 868.
• ⁷
  Zowtiak, J. M.; Bartholomew, C. H.; J. Catal 1983, 83, 107.
• ⁸
  Reuel, R. C.; Bartholomew, C. H.; J. Catal.1984, 85, 63.
• ⁹
  Rane, S.; Borg, O.; Yang, J.; Rytter, E.; Holmen, A.; Appl. Catal., A2010, 388, 160.
• ¹⁰
  Sexton, B. A.; Hughes, A. E.; Turney, T. W.; J. Catal.1986, 97, 390.
• ¹¹
  Carta, G.; Casarin, M.; El Habra, N.; Natali, M.; Rossetto, G.; Sada, C.; Tondello, E.; Zanella, P.; Electrochim. Acta2005, 50, 4592.
• ¹²
  Hilmen, A. M.; Schanke, D.; Hanssen, K. F.; Holmen, A.; Appl. Catal., A1999, 186, 169.
• ¹³
  Lee, J. C.; Kang, H. J.; Kim, K. J.; Kim, Y. S.; Moon, D. W.; Surf. Sci.1976, 324, 338.
• ¹⁴
  Kerkhof, F. P. J.; Moulijn, J. A.; J. Phys. Chem.1979, 83, 1612.
• ¹⁵
  Penn, D. R.; J. Electron Spectrosc. Relat. Phenom.1976, 9, 294.
• ¹⁶
  Liotta, L. F.; Pantaleo, G.; Macaluso, A.; Di Carlo, G.; Deganello, G.; Appl. Catal., A2003, 245, 167.
• ¹⁷
  Jongsomjit, B.; Panpranot, J.; Goodwin, J. G.; J. Catal 2001, 204, 98.
• ¹⁸
  Van Steen, E.; Sewell, G. S.; Makhothe, R. A.; Micklethwaite, C.; Manstein, H.; De Lange, M.; O'Connor, C. T.; J. Catal 1996, 162, 220.
• ¹⁹
  Storsaeter, S.; Totdal, B.; Walmsley, J. C.; Tanem, B. S.; Holmen, A.; J. Catal 2005, 236, 139.
• ²⁰
  Zhang, J. L.; Chen, J. G.; Ren, J.; Sun, Y. H.; Appl. Catal., A2003, 243, 121.
• ²¹
  Chin, R. L.; Hercules, D. M.; J. Phys. Chem.1982, 86, 360.
• ²²
  Nagaraju, N.; Fonseca, A.; Konya, Z.; Nagy, J. B.; J. Mol. Catal. A: Chem.2002, 181, 57.
• ²³
  Chai, S. P.; Zein, S. H. S.; Mohamed, A. R.; Chem. Phys. Lett 2006, 426, 345.
• ²⁴
  Dai, H.; Surf. Sci 2002, 500, 218.
• ²⁵
  Stranick, M. A.; Houalla, M.; Hercules, D. M.; J. Catal 1987, 103, 151.
• ²⁶
  Grimm, D.; Gruneis, A.; Kramberger, C.; Chem. Phys. Lett.2006, 428, 416.
• ²⁷
  Saito, R.; Fujita, M.; Dresselhaus, G.; Dresselhaus, M. S.; Appl. Phys. Lett 1992, 60, 2204.
• ²⁸
  Alvarez, W. E.; Pompeo, F.; Herrera, J. E.; Balzano, L.; Resasco, D. E.; Chem. Mater 2002, 14, 1853.

Publication Dates

• Publication in this collection
  Dec 2014

History

• Received
  5 Aug 2014
• Accepted
  17 Oct 2014