Synthesis of MWCNTs by chemical vapor deposition of methane using FeMo/MgO catalyst: role of hydrogen and kinetic study

This study aims to investigate the role of hydrogen on CNTs synthesis and kinetics of CNTs formation. The CNTs were synthesized by catalytic chemical vapor deposition of methane over FeMo/MgO catalyst. The experimental results revealed that hydrogen plays an important role in the structural changes of catalyst during the pre-reduction process. The catalyst structure fully transformed into metallic FeMo phases, resulting in an increased yield of 5 folds higher than those of the non-reduced catalyst. However, the slightly larger diameter and lower crystallinity ratio of CNTs was obtained. The hydrogen co-feeding during the synthesis can slightly increase the CNTs yield. After achieving the optimum amount of hydrogen addition, further increase in hydrogen would inhibit the methane decomposition, resulting in lower product yield. The hydrogenation of carbon to methane was proceeded in hydrogen co-feed process. However, the hydrogenation was non-selective to allotropes of carbon. Therefore, the addition of hydrogen would not benefit neither maintaining the catalyst stability nor improving the crystallinity of the CNT products. The kinetic model of CNTs formation, derived from the two types of active site of dissociative adsorption of methane, corresponded well to the experimental results. The rate of CNTs formation greatly increases with the partial pressure of methane but decreases when saturation is exceeded. The activation energy was found to be 13.22 kJ mol−1, showing the rate controlling step to be in the process of mass transfer.


CNTs synthesis
CNTs were synthesized by the catalytic chemical vapor deposition (CCVD) method using methane as a carbon source.The schematic diagram was shown in Fig. 2. Firstly, 0.5 g of FeMo/MgO catalyst was put in a quartz boat, then placed in the quartz tube reactor center.This CCVD process started with N 2 fed into the reactor and heated up from room temperature until it reached desired temperature by linearly increasing at a rate of 10 °C min −1 .
The total gas flow rate was controlled with following each condition, which calibrated by soap film meter.The effluent gas was collected and analyzed with Gas Chromatography (GC-TCD 8A, Shimadzu).Conditions for GC analysis were set as follows (3 mm ∅ × 2 m column, INJ/DET: 120 °C, COL: 100 °C, active carbon packing: 0.2-0.25 mm, and He carrier: 40 ml min −1 ).Until the end of the reaction (180 min), the system was cooled to room temperature in N 2 gas flow.The CNTs were formed at the surface of the catalyst in the quartz boat.
The catalytic performance in terms of CH 4 conversion, Carbon yield, g-CNTs/g-catalysts, and calculated H 2 flow rate were calculated according to Eqs. ( 1)-(4), respectively. (1) (2) Carbon yield(%) = product weight g − catalyst weight g carbon feed g × 100% www.nature.com/scientificreports/ The kinetics investigation in "Kinetic study of CNTs synthesis" section will perform experiment with H 2 pre-reduction process (R-woH 2 ).The product yield used to calculate yield rate as an Eq.(5).The reduced catalyst weight after the reduction process will be 0.75 times to the initial catalyst weight.Rate of reaction will be calculated by Eq. (6)   (3) g − CNTs g − catalysts = product g − catalyst weight g catalyst weight g (4) Calculated H 2 flow rate(ml/min) = F outlet −F N2 −F unreacted, CH 4 (5)  Product yield g = g product − g reduced−catalyst ; g reduced−catalyst = 0.75xg catalyst

Characterization
The study of morphology, internal, and external structure was examined using imaging process.The external structure of as-prepared and purified product was observed by Field Emission Scanning Electron Microscopes (FE-SEM, HITACHI SU-8010).The internal structure, type of CNTs, number of walled, CNTs growth mechanism (Tip, based growth), CNTs and catalyst diameter size were observed by Transmission Electron Microscopes (TEM, JEOL JEM-2100 Plus).Sample preparation for TEM analyzed was prepared by following step; the product was dispersed in ethanol (99.5v/v%,Sigma-Aldrich) and then dropped onto a 300 mesh of copper grid coated with carbon film.As-prepared CNTs and catalysts diameter size were measured from SEM, and TEM images by using ImageJ software.Crystallinity and structure of CNTs and catalysts were analyzed for study of changing in the sample, because of the controlled processes.X-ray diffraction (XRD) with the angle range scan 5°-80° and TEM selected area electron diffraction (SAED) were used to specify crystallinity, phase change, and crystal size.
The crystallinity of the synthesized carbon products, which validated with qualitative analysis of the proportion between crystalline carbon and amorphous carbon was carried out using 532 nm laser light source Raman spectroscopy (Raman NT-MDT model: NTEGR-SPECTRA).The purity of as-prepared product was measured from the thermo-gravimetric analysis (TGA), by observing the weight loss in expected temperature range.

Effect of the presence of H 2
In the study roles of H 2 , there are two variables that need to be studied which are, the pre-reduction of catalyst and the H The catalytic activity was examined over CCVD for synthesis of CNTs process.From the experiment, FeMo/ MgO catalyst is capable of converting CH 4 in all processes as shown by methane conversion in Fig. 3.
Methane conversions decreased over time because of catalyst deactivation.The formation of product CNTs obstructed the active surface of catalysts.The catalyst performance depends on methane conversion, deactivated rate, and quantity of final product.It was found that R-woH 2 was the most effective process with a 3.41 g final product followed with R-wH 2 (3.33 g), nR-wH 2 (1.19 g), and nR-woH 2 (0.93 g), respectively.The thermal stability of the product was observed with thermo-gravimetric analysis (TGA) in air atmosphere, as shown in Fig. 4.
From Fig. 4, it was found that product begins to decompose at temperatures of 600 °C and above.Decomposition curves showed similar decomposed characteristics or referring to the same group of products.Residuals from decomposition under air atmosphere are catalyst and supported materials which are highly thermal stability than carbon product, therefore we can get carbon content from TGA.The percentage carbon content corresponds to the yield.Products from pre-reduction of catalyst process (R-woH 2 , R-wH 2 ) have higher carbon content than non-reduced processes.
The morphology of as-prepared products from various processes was investigated by image processing through scanning electron microscope (SEM) as shown in Fig. 5. Dense CNTs were found in all SEM images included with some amorphous carbon.Considering the CNTs products, the diameter of MWCNTs was measured from over 200 samples at 10,000×.magnification combined with over 100 samples at 50,000×.magnification by using ImageJ software to produce a diameter size distribution of as-prepared CNTs from each process.
The diameter size distribution of CNTs represents non-uniform product, ranging from tiny CNTs (smaller than 5 nm) to huge CNTs (larger than 100 nm), as a result of improper catalyst preparation method, thus affecting the active size which is expressed as the diameter of CNTs 36 .Considering the mean diameter, CNTs from nonreduced processes: nR-woH 2 and nR-wH 2 were approximately the same sizes (29.29 and 32.72 nm) meanwhile, the reduced process: R-woH 2 and R-wH 2 have similar in diameter (43.25 and 45.41 nm) but were larger than CNTs-nR processes.
Qualitative analysis of the proportion between crystalline carbon and amorphous carbon was carried out using 532 nm laser light source Raman spectroscopy.Raman is used to determine proportions between (6) r CNTs gg −1 cat.s −1 = product yield g g reduced−catalyst × reaction time(S)   3.
From the Raman analysis, it was found that the products formed by non-reduced processes have higher I G /I D ratio than reduced processes which approximately are 3.1 and 2.5, respectively.Observed that the addition of H 2 during the formation of CNTs did not cause a difference in the crystallinity ratio (more detailed in  "Hydrogenation reaction" section).Therefore, there was a possibility that the crystallinity ratio was also controlled by the catalyst structure.
In the study of H 2 roles, the samples were clearly divided into two groups which are non-reduced processes and reduced process.Non-reduced process (nR-woH 2 and nR-wH 2 ) produces products with smaller diameters and higher crystallinity ratios than reduced processes (R-woH 2 and R-wH 2 ), but reduced processes have a higher yield than 5 times (~ 1 to ~ 5) of non-reduced process.
The structure of the catalyst and as-prepared product in each process were investigated by XRD method to describe the mechanisms that cause changes in CNTs.Firstly, the H 2 -temperature programmed reduction (H 2 -TPR) analysis was used to investigate H 2 consumption during temperature changes, which is a similar experimental to pre-reduction of catalyst process.In the reduced process, it was confirmed by H 2 -TPR that prereduction of catalyst from room temperature to 900 °C would allow the catalyst to remain structurally stable or without structural changes in the presence of hydrogen in the system 3 , as shown in Figure S1, that H 2 is not consumed at 900 °C and above.
XRD patterns of catalysts consisted of calcined MgO, calcined FeMo/MgO, and reduced FeMo/MgO are shown in Fig. 7.The catalyst phases were identified by the diffraction peaks.The diffraction peaks of magnesium oxide (MgO, PDF 01-071-1176) were observed in Fig. 7a which were cubic structures.Figure 7b exhibited the   diffraction peaks of MgO impregnated with iron (Fe) and molybdenum (Mo), where only monoclinic MgMo 4 at 23.2° (021), and 26.3° (220) 37 was found.Xu et al. 38 reported an XRD analysis of calcined FeMo/MgO, revealing the structure of MgO, Fe 3 O 4 , and MgMo 4 .The results of the xrd diffraction were similarly blunt same as this work.XRD patterns of reduced FeMo/MgO was show in Fig. 7c, tetragonal FeMo exhibited the diffraction peak at 37.3° (002), 44.6° (411), and 64.7° (223) 39 .Figure 7 found that the calcination process did not detect Fe-Mo interaction, FeMo was found after the catalytic reduction process.Therefore, the reduction of catalyst process changes the catalyst structure from the oxide form to metallic structure and establishes an interaction between iron and molybdenum.
Comparison of product structure between reduced FeMo/MgO, non-reduced CNTs (nR-woH 2 ), and reduced CNTs (R-woH 2 ) is shown in Fig. 8. FeMo undetectable after CNTs growth from both of non-reduced and reduced processes, which may cause by carbon obscuring or active metal was encapsulated within CNTs.Diffraction patterns of hexagonal carbon or carbon nanotubes was found at 25.9° (002) 40 , which shown in Fig. 8b and c.
The XRD patterns of as-prepared CNTs are shown in Fig. 9.It was found that the product characteristics are divided into two groups in the same way as the properties of CNTs, which are non-reduced processes and reduced processes.
The details of the XRD analysis results in the 30°-60° and 68°-80° ranges are shown in Fig. 10A and B, respectively.The diffraction pattern exhibited iron carbide (Fe 3 C) that can be found in all processes.Meanwhile, diffraction peak of molybdenum carbide (Mo 2 C) is discovered in non-reduced processes.Therefore, in a non-reduced process, the catalyst is converted from oxide to carbide form.In addition, iron, and molybdenum interacted poorly, resulting in the formation of molybdenum carbide structures.These results show why synthesized CNTs by non-reduced and reduced processes differ.
Several previous studies have mentioned similar behaviors with Fe catalysts.He et al. 23 reported that the carbide catalyst (Fe 3 C) will make CNTs with a smaller diameter but lower yield than the metallic catalyst (α-Fe).In this work, we found that non-reduced processes will be made the carbide catalyst, but reduced processes will be made metallic catalyst (FeMo).This is why products from the non-reduced and reduced processes are different.
The carbide formation of catalyst is described by Behr et al. 15 .They studied the H 2 -to-CH 4 ratio in the feed gas.Carbide formation will occur when lower H 2 -to -CH 4 ratio, while the metallic formation will occur when a higher H 2 -to-CH 4 ratio.Reduced processes in this work occur in H 2 and N 2 atmosphere, in the meaning there is no carbon (CH 4 ) in the system, that made a FeMo metallic, but for non-reduced processes, catalyst will be reduced by H 2 gas dissociated from CH 4 , because CH 4 can be decomposed by the oxide form of the catalyst 41 .The reaction product hydrogen gas may make the reduction form of the catalyst with presence of carbon atoms on catalyst surface that caused a carbide catalyst.
Therefore, from the study of H 2 roles, it was found that the characteristics of CNTs, whether yield, diameter, and proportion of crystallinity, are determined by the structure of catalyst that was controlled by the prereduction process.Methane conversion was investigated and H 2 flow rate was calculated by Eq. 4 then plotted against reaction time as shown in Fig. 11.As-prepared CNTs were measured for average diameter and summarizing the results with other information in Fig. 12 and Table 4. Methane conversion in each H 2 concentration was slightly different as shown in Fig. 11a, where the CH 4 :H 2 ratio was 1:1 with the highest methane conversion.In Fig. 11b, exhibited H 2 flow rate that is higher than amount of H 2 fed into the system.Since H 2 is one of the products resulting from the decomposition of methane, it was found that H 2 flow rate decreases over time corresponding to the reduced methane conversion.
Figure 12 shows the yield of the produce (g-CNTs/g-catalyst) with an optimum point of 1:1, corresponding to the conversion of methane.When the H 2 content is increased above 1:1, the yield is less.Observing the flow rate of H 2 , found that there is high amount of H 2 in system.The study of Kashiwaya Y., and Watanabe M. reported that methane decomposition is reversible reaction in gas phase before CNTs growth 42 .Therefore, according to Le Chatelier's principle, when the product quantity (H 2 ) increases, will prevent the dissociation of methane, resulting in fewer CNTs 43,44 .Additionally, the addition of the proper amount of H 2 allows catalyst to restructure that ready for the formation of higher CNTs.Observed from H 2 -free process (nR-woH 2 ) with the lowest methane conversion (~ 8%) at 5 min, then the conversion increased.Because initially the catalyst was in an oxide form with lower methane conversion performance, and when partially H 2 was formed, the catalyst changed its structure and thus the conversion increased.
The results of the study on adjustment concentration of H 2 gas, when adding H 2 , yield was higher but decreased when excess H 2 .Therefore, adding H 2 to CNTs product has an optimal point, that consistent with various research [12][13][14][15][16][17][18] .
Measure and establish the CNTs diameter size distribution from the study of effects of H 2 concentration as shown in Fig. 12 and Table 4, found that as-prepared CNTs from this work are highly distributed.As a result, it was unable to confirm the results of the study of the diameter changes from the adjustment of the H 2 concentration.
However, several studies indicate that H 2 addition helps to eliminate amorphous carbon as well as maintain catalyst stability.To study about this issue, the study of catalytic hydrogenation reaction was conducted in the next section.

Hydrogenation reaction
This section examined the catalytic hydrogenation reaction that was thought to play an important role in the synthesis of CNTs, by investigating whether this reaction can help stabilize the catalyst during the synthesis of CNTs, included with an improvement of CNTs quality.Piedigrosso et al. 11 reported that, CNTs are capable of catalytic hydrogenation, but there is no selective eliminate of solid carbon, either crystalline carbon or amorphous carbon.Catalytic hydrogenation may enhance the catalyst performance by removing the carbon covering the catalyst surface, which caused the deactivation of catalyst.In this work, an experiment was conducted to study the synthesis of CNTs in reduced-process, by comparing between processes with and without H 2 feed (R-woH 2 and R-wH 2 ) with a reaction time of 5 h (300 min) to study catalyst stability of both processes (Figure S2).It was found that the conversion of methane continues to decrease at the same rate in both processes.Comparison the product weight difference between two processes, at 3 h, the difference was about 0.07 g, but at 5 h, the difference was about 0.26 g.Therefore, it seems that addition of H 2 does not keep catalyst stability, but also reduces the weight of the product.
The catalytic hydrogenation reaction was tested to prove these issues.As-prepared CNTs from R-woH 2 and R-wH 2 process were tested for catalytic hydrogenation reaction at 900 °C with flow rates of 100 and 50 ml min −1 of H 2 and N 2 , respectively.Effluent gas, product weight, and crystalline ratio were analyzed to identify the behavior that occurred.Figure 13 exhibits flow rate of CH 4 and CO released from the system during the catalytic hydrogenation reaction test.It also shows the calculated quantity of CNTs consumed as a %yield.From the experiment, it was found that CO was initially formed.Methane occurred throughout the experiment, with the flow rate decreasing slightly over time.After the experiment, approximately 3% of CNTs were consumed.In addition, CH 4 and CO generated rate have no different for each process, although these two samples are expected to have different     incidences, because R-wH 2 has been occurred in catalytic hydrogenation (during the formation of CNTs) for longer than R-woH 2 .If catalytic hydrogenation is a selective reaction, a difference will occur in some way.
The thermal stability of the product comparing between As-received CNTs (from process R-woH 2 and R-wH 2 ) and CNTs after process Catalytic hydrogenation reaction (CHT) is shown in Fig. 14, and the average crystallinity ratio by Raman characteristic is shown in Table 5.From the thermal decomposition in Fig. 14, it was found that all samples were not different, with similar carbon content and decomposed at the same temperature range.Consider the proportion of crystallinity.Raman analysis has shown no difference in I G /I D ratio with an average of 2.6.
Therefore, from the study of the catalytic hydrogenation reaction.It can be concluded that the catalytic hydrogenation reaction occurs during the synthesis of CNTs, whereas CNTs are used as a precursor to form methane and carbon monoxide.This reaction was not selective to eliminate either crystalline or amorphous carbon.As a result, it does not improve the stability of the catalyst, by etching carbon from surface of catalyst.The addition of H 2 to the system will not enhance the properties of CNTs and produce less product, because of the catalytic hydrogenation reaction and the shifted of reaction equilibrium.
From the study of the role of H 2 in "Effect of the presence of H2" section, it was concluded that an important role of H 2 is to determine the catalyst structure by H 2 pre-reduction of catalyst before CNTs formation.Whereas the addition of H 2 during the formation of CNTs was not found to have a beneficial effect on the synthesis of CNTs and resulted in less product yield.

Kinetic study of CNTs synthesis
The aim of the kinetic study of CNTs synthesis was to investigate the mechanisms occurring during the synthesis of CNTs, with the hope that the data could be used for further process scale-up.Data collection for use in kinetic studies in CNTs synthesis is quite difficult because solids form while reaction that made an unsteady state reaction.Research works have chosen a variety methods to investigated reaction rate, such as measuring the height of VACNTs 45 , analysis of hydrogen output gas 46 , or batch analysis by collecting data on changing weight of the catalyst bed 47,48 .In this work, batch analysis was used.The reaction rate in each process was collected as shown in Fig. 15a.The rate of product yield was calculated according to the Eq. ( 6), with units of gram product divided by grams of catalyst and reaction time in second.A dramatic decrease in yield rate can be observed in all processes.From the study in "Roles of hydrogen" section it was found that R-woH 2 process had the highest yield and was able to reduce the complexity of analysis by removing H 2 feeding.Therefore, this process was chosen for further kinetic studies with 30 min reaction time.The partial pressure of CH 4 in feed gas was varied to identify the reaction order and effects of CH 4 concentration.The experimental data for kinetic study was shown in Table S1.Partial pressure of CH 4 was conducted by varying of CH 4 flow rate.
Yadav et al. 49 derived model 1 and 2 by assuming dissociative adsorption of methane followed by removal of hydrogen from the adsorbed methyl group as the rate determining step.The reaction step was considered as irreversible reaction and the difference between model 1 and 2 is the one type of active site (x) or two types (x and y).Pseudo first order reaction kinetic, kinetic model 1 and model 2 were used to compare for this reaction, leading to the following rate Eqs.(7-9).The rate constants are given by k ' , k 1 and k 2 , respectively.To estimate the parameters k, Eqs.(7-9) can be linearized and parameters were estimated from the slope and intercepts.
The experimental and the simulated values from models were plotted as shown in Fig. 15b.The partial pressure of methane equal to 1 does not consider as it is a pure methane system which is inconsistent with the other proportions.The rate of formation of CNTs increased with an increase in the partial pressure of methane, then reaches as saturation level followed by decrease in the rate at higher partial pressures of methane.The value of the rate constant from each model was shown in Table 6.Yadav et al. resulted as well.It can be seen that model 1 predicts the reaction rate same as model 2 which is better than 1st order reaction over a covered condition in this work.These results show that the decomposition of methane into CNTs and hydrogen is might be 1st order reaction but, model 1 and model 2 can predict more accurately.
The effect of temperature was studied in the range of 850-1000 °C with partial pressure of methane equal to 0.24 atm.(50 ml min −1 ).It was observed that the rate of CNTs formation increased with temperature until higher than 950 °C because of catalyst begin to be agglomerated that decreased catalyst performance, as shown in Fig. 15c.
To estimate the activation energy (E a ), Arrhenius plot is needed.Then the rate of reaction (lnR CNTs ) was plotted versus T −1 as shown in Fig. 15d.The rate of CNTs formation at 1000 °C was not considered for E a estimation, as agglomeration of active phase occurred.
The activation energy was found to be 13.22 kJ mol −1 , compared to the other tasks in summarize table of Yadav et al. 49 work, our E a was much lower.It may be in a range of mass transfer limit 50 , consistent with an assumption of pseudo first order reaction that used to describe mass transfer limit the reaction rate.Additionally, our experiment found the behavior of the CNTs formation rapidly occurs, and a compaction of the bed layer will occur, which may hinder the formation of CNTs, showing that this experimental process undesignated for study mass transfer limit process.Therefore, it is possible that E a has a lower value compared to other works.Moreover, E a determinations should be carried out in reactions with low conversion and reaction temperatures to get accurate and reasonable values 51,52 . (7)

Conclusion
In this study, the role of H 2 on CNTs synthesis was investigated by means of pre-reduction of catalyst and cofeeding during CNTs synthesis.For FeMo/MgO catalyst, the reduced process could produce CNTs with the largest yield ascribed to the presence of highly active FeMo metallic phases.Meanwhile, the Mo 2 C phase was formed in the non-reduced process, resulting in the formation of CNTs with smaller diameter and higher crystallinity.The co-feeding of H 2 during CNTs synthesis in the non-reduced process can increase the yield of CNTs formation.However, the yield became lower when H 2 exceeded the equilibrium of reaction.In addition, the presence of H 2 during the CNTs synthesis would promote the catalytic hydrogenation of carbon to methane.This reaction was not selective for elimination of amorphous carbon.As a result, the co-feeding of H 2 would not improve neither the CNTs crystallinity nor the stability of the catalyst.The kinetic study of CNTs synthesis shows that methane decomposition to CNTs and hydrogen is a pseudo first order reaction with mass transfer as rate-controlling step.
The rate of CNTs formation increases with an increased methane partial pressure.The activation energy derived from the experimental data was found to be relatively low at 13.22 kJ mol −1 .In sum, the synthesis of MWCNTs by CCVD of methane using FeMo/MgO catalyst can be achieved under H 2 free condition.The characteristics of as-grown CNTs would greatly depend on the structure of the catalyst.The adding of hydrogen shows an important role in changing the catalyst structure during the pre-reduction of catalyst.
graphitize carbon and disorder or amorphous carbon by the relative height of G band (~ 1580 cm −1 ) and D band (~ 1350 cm −1 ) that represent graphitic and disorder carbon, respectively.The G/D peak intensity ratios (I G /I D ) of a carbonaceous product were investigated more than 5 points for each process to confirm the results of the analysis and prevent false sampling.Examples of the Raman analysis result are shown in Fig. 6.Average I G /I D ratio and other detailed data for the study of overview H 2 roles are shown in Table

Figure 3 .
Figure 3. Catalytic activity through (solid line) methane conversion and (dotted line) product weight.

Figure 4 .
Figure 4. TGA of as-prepared CNTs by comparing each H 2 processes.

Figure 11 .
Figure 11.Effect of H 2 concentration on (a) CH 4 conversion (b) calculated H 2 flow rate over FeMo/MgO catalyst at 900 °C.

Figure 12 .
Figure 12. g-CNTs/g-catalyst, and average CNTs diameter as a function of CH 4 :H 2 flow rate ratio.Error bars are the standard deviation for each sample and is a measure of the diameter distribution.

Figure 13 .
Figure 13.Flow rate of CH 4 and CO, and %yield as a funtion of catalytic hydrogenation reaction time.

Figure 14 .
Figure 14.TGA of CNTs for study of catalytic hydrogenation reaction.

Figure 15 .Table 6 . 1 k 2 × 2 Model 2 k 2 ×
Figure 15.(a) The yield rate of CNTs synthesis at 900 o C, (b) Effect of partial pressure of methane on the yield rate of CNTs formation at 900 o C with line indicating the calculated rates by kinetic model, (c) Effect of temperature on the CNTs yield rate, and (d) The Arrhenius plot for estimation of the activation energy.

Table 1 .
Condition for synthesis of CNTs.

Table 3 .
Summary data for the study of overview H 2 roles.

Table 4 .
The summary data for study of effects of H 2 concentration.

Table 5 .
Purity and I G /I D ratio for study of catalytic hydrogenation reaction.