Understanding the role of post-CCVD synthetic impurities, functional groups and functionalization-based oxidation debris on the behaviour of carbon nanotubes as a catalyst support in cyclohexene hydrogenation over Pd nanoparticles

Abstract

A new method has been developed to separately study the effects of (i) impurities resulting from the catalytic chemical vapor deposition synthesis, (ii) the attached functional groups and (iii) the oxidation debris on the properties of carboxyl-functionalized multiwall carbon nanotubes (CNTs) by incorporating a Soxhlet-extractor enhanced acetone washing process into the synthesis method. Pd nanoparticles supported on the carbon nanotubes were investigated in the hydrogenation of cyclohexene to cyclohexane. Despite the fact that the specific surface area and the Pd dispersion were both low, the Pd/CNT catalyst with post-synthetic impurities showed ∼20 times higher catalytic activity compared to functional group free, acetone-washed samples. Meanwhile oxidation debris originating from the functionalization was found to affect both the specific surface area and the G/D ratio obtained from Raman spectra, it had slight effect on the size of the supported Pd nanoparticles or the catalytic activity. On the other hand, functional groups have a significant effect on the catalytic activity without influencing the specific surface area, the G/D ratio or the Pd nanoparticle dispersion.

---

1. Introduction

Carbon nanotubes (CNTs) have been the subject of considerable scientific interest in the past few decades, especially in the field of heterogeneous catalysis. Nanoparticles of different types and sizes supported on CNTs have been tested in many catalytic reactions, such as hydrogenation, dehydrogenation, and oxidation.^1–5 The most common methods of carbon nanotube synthesis are based on the catalytic chemical vapor deposition (CCVD) process. This method can result in amorphous carbon based impurities⁶ that may affect the catalytic performance.

The effect of carbon nanotube oxidation on the sorption of both organic and inorganic substances,^7–10 the size of supported nanoparticles^11,12 and the length of carbon nanotubes^13–15 has also been extensively studied. In the case of nanotube length, it has been found that the extended oxidation time reduces nanotube length significantly and the most prominent shortening occurs in the first few hours of oxidation resulted in several implications on catalytic reactions or adsorption performance.

Some oxidation techniques can produce oxidation debris and functional groups on the carbon nanotube surfaces, which can influence chemical and biological properties. During intensive oxidation, holes can form in the individual carbon layers and can grow up to a point where they are interconnecting and leave small parts of the outer graphitic layer separate from the main layer. These small oxidized graphene “islands” then get peeled off to either form debris on the surface by aggregation or be oxidized completely to CO₂.^16,17 The formation and decomposition of oxygen containing functional groups on carbon nanotubes was also studied using different acids.^18–20 It has been found that different types of functional groups, like carboxyl, carbonyl, quinone, aldehyde, etc. form on the surface which decompose at different temperatures. At temperatures ranging from 150 to 450 °C, carboxylic and phenolic hydroxyl groups were found to dehydrate. Carboxylic groups were also found to decompose at this temperature range, while the previously formed anhydrides and other groups like ketones or lactones were found to be thermally more stable. X-ray photoelectron spectroscopy data suggests that functional groups still remain on the surface even after treatments at temperatures as high as 720 °C.

Zhang and co-workers investigated how defects are generated in single walled carbon nanotubes subjected to different oxidants. It has been found that after the initial attack of the oxidant on existing active sites (e.g. –CH, –CH₂ groups, Stone–Wales defects) an electrophilic addition begins at hexatomic–hexatomic boundries producing new active sites or rather, new defects. The authors consider this to be a defect generating step, which is then accompanied by a defect consuming step, where the graphitic structure around existing active sites is destroyed during intense oxidation.²¹

The effect of removal of such debris has been studied to some extent^22–26 already. The most common removal methods are based on washing with aqueous solutions under neutral or basic conditions. However, the basic washing solution can react with the surface carboxylic groups of the carbon nanotubes, while water will not necessarily dissolve all organic residues and consequently, the surface is either altered or not cleaned properly. The use of basic solutions also arises the question of introducing unnecessary alkali elements onto the surface which are known to have promoting effects on certain catalytic reactions, which effect can present itself as an artifact under certain test conditions.^27–29

In this study, we developed a Soxhlet-extractor washing process with acetone to remove all forms of synthesis and oxidation debris from CNT surfaces. This process allowed us to separately study the effects of post-synthetic (post-CCVD) impurities, wet chemical oxidation-based functional groups and attendant oxidation debris on the various chemical and physical properties of carbon nanotubes without modifying the surface through chemical reactions or introduction of foreign elements. In order to follow these effects, CCVD synthesized multiwall carbon nanotubes have been oxidized in concentrated nitric acid for 4–24 h and washed in acetone using a Soxhlet-extractor. The order of oxidation and washing steps were altered, which allowed revealing the effects of post-CCVD impurities, functional groups and oxidation debris separately by the process of elimination. The resulting changes in specific surface area and general carbon structure were followed by nitrogen porosimetry and Raman spectroscopy. Finally, the functionalized carbon nanotubes were impregnated with Pd nanoparticles and catalytic hydrogenation of cyclohexene was performed on the Pd/CNT catalysts. The duration of oxidation and the order of washing have highly affected the physical properties of the carbon nanotubes as well as the size of the supported Pd nanoparticles and the catalytic activity, which suggests a complex role of post-CCVD impurities, functional groups and oxidation debris.

2. Materials and methods

2.1 Synthetic procedures

Multiwall carbon nanotubes (MWCNT) were prepared in our laboratory by the catalytic chemical vapor deposition (CCVD) method described elsewhere.³⁰ The as-prepared MWCNT samples were modified in different washing and functionalization processes as described as follows:

The oxidation of MWCNTs was performed by thermally assisted oxidation, where fixed amounts (4–4 g) of MWCNTs were refluxed in cc. HNO₃ solution (500 mL, 65 wt%) for 4, 8, 12, 16, 20 and 24 h to generate oxygen-containing surface functional groups. The products were washed with distilled water to neutral pH and dried overnight at 80 °C in air. Non-oxidized samples were also part of the investigations.

In one series, each sample containing 4 g of functionalized carbon nanotubes with different oxidation durations were washed with 500 mL of acetone in a Soxhlet-extractor for 48 h and dried again overnight at 80 °C. These samples are denoted as “After Washed” carbon nanotubes (CNT-AW). Acetone was chosen as washing agent for its low boiling point, which ensures that washed off impurities do not recirculate during the washing process.

In another series, the functionalized carbon nanotubes were washed with the same procedure as described above before the oxidation-assisted functionalization. These samples are denoted as “Before Washed” carbon nanotubes (CNT-BW).

The effect of washing was compared to functionalized carbon nanotubes not subjected to any washing process. These samples are denoted as “Not Washed” carbon nanotubes (CNT-NW).

Palladium nanoparticles were impregnated onto the surface of the different MWCNT samples by the well-known wet impregnation method using toluene as the medium and palladium(II)-acetate (Sigma-Aldrich) as the palladium source. In a typical procedure, 22 mg of Pd(OAc)₂ was dissolved in 100 mL toluene, then 200 mg CNT was added to the solution and the mixture was sonicated for 15 min in an ultrasonic bath (80 W) and stirred for 24 h at room temperature. The samples were then centrifuged at 3200 rpm and dried in air at 80 °C overnight. The as-prepared samples were finally annealed at 185 °C for 2 h followed by 380 °C for 1 h in nitrogen atmosphere, where the metal salt completely decomposed to yield metallic Pd nanoparticles on the carbon nanotube surface.³¹ Pd-impregnated samples are denoted as Pd-CNT-AW, Pd-CNT-BW and Pd-CNT-NW. A simplified graphical schematic for the various preparation steps is presented in Fig. 1.

Fig. 1 A graphical overview of the effects of various sample preparation steps on the surface environment of carbon nanotubes.

2.2 Characterization

2.2.1. Raman spectroscopy. Raman spectra were collected at 4 cm⁻¹ resolution for 3 min using a Thermo Scientific Raman DXR microscope with a 532 nm laser excitation operated at 5 mW. Nitrogen adsorption data were collected on a Quantachrome NOVA 3200 instrument. All samples were degassed for 1.5 h in vacuum at 380 °C before measurement.

2.2.2. Transmission electron microscopy. Transmission electron microscopy (TEM) was performed using a FEI Tecnai G² 20 X-TWIN microscope with a tungsten cathode operated at 200 kV.

2.2.3. X-ray diffraction. Powder X-ray diffraction (XRD) patterns were collected in a Rigaku Miniflex II desktop diffractometer using Cu K_α irradiation at a scan speed of 4° min⁻¹ and 0.25° min⁻¹ in the ranges 2θ = 5–90° and 2θ = 35–50°, respectively.

2.2.4. Energy dispersive X-ray spectroscopy. To determine the Pd concentration, several Energy Dispersive X-ray Spectroscopy (EDS) measurements were conducted on each sample with a Röntec QX2 detector mounted on a Hitachi S-4700 scanning electron microscope.

2.3 Cyclohexene hydrogenation

Catalytic hydrogenation of cyclohexene to cyclohexane was performed in a continuous flow reactor system at 40 °C, where the ratio of cyclohexene : hydrogen : nitrogen was 1 : 10 : 90 with a total flow rate of 101 mL min⁻¹ at 1 bar. Product analysis was performed on-line with an Agilent 6820 gas chromatograph using a flame ionization detector. The aforementioned catalytic model reaction was chosen for its nature not to produce any by-products under the given experimental conditions. The method used for calculating TOF is provided in ESI.†

3. Results and discussion

3.1 Raman spectroscopy

Raman spectra were collected on all three sample series (CNT-AW, CNT-BW and CNT-NW samples). Raman spectra are presented in Fig. S3.† The intensity ratio of the well-known Raman D and G bands at ∼1349 and 1586 cm⁻¹, respectively, was examined to reveal possible structural changes of carbon nanotubes with the functionalization time (Fig. 2). The graphitic or G band is attributed to the tangential lattice vibrations of sp² bonded carbon atoms. Meanwhile, the intensity of the disorder induced D band is generally associated with the amount of defects, or more precisely the amount of sp³ hybridized phonon scattering sites. The relatively low initial G/D value of 0.95 for non-oxidized samples can be attributed to the moderate growth temperature of nanotubes, as higher growth temperatures are known to result in nanotubes with better quality graphitic structures.³²

Fig. 2 Raman G/D band ratios of multiwall carbon nanotube samples after different functionalization durations (4–24 h) in the case of “Non Washed”, “Before Washed” and “After Washed” samples (lines are guide to the eye).

In case of CNT-NW, the G/D ratio decreased from 0.95 to 0.9 after 4 h of oxidation then further to the final value of 0.71 (25% drop) in the next 20 h. This is in agreement with previous studies where nitric acid was used for the oxidation of carbon nanotubes. Interestingly, both the CNT-AW and the CNT-BW series show ∼20% decrease in G/D ratio after 4 h oxidation time already. Values of CNT-AW's are slowly decreasing from 0.78 to 0.76 from 4 h to 16 h, reaching 0.70 at even longer functionalization times. After the ∼20% drop in the first 4 hours, the CNT-BW samples show a slow G/D ratio decrease from 0.75 to 0.70 between 4–16. The G/D ratio is independent of the washing process if functionalization is performed for longer than 20 h.

A clearly distinguishable G/D ratio decrease is visible in both the CNT-NW and the CNT-BW sample series between 4 and 16 h of oxidation. This drop can be attributed to the continuously formed oxidation debris and functional groups created by the nitric acid oxidation, as no washing process was used after functionalization in these series.

However, the G/D ratios of the CNT-NW are always higher than those of the CNT-BW samples, which can be attributed to the application of the washing process before oxidation in the case of CNT-BW. As washing applied before oxidation can only remove already present impurities, namely the post-CCVD ones, the higher G/D values of CNT-NW compared to CNT-BW indicate that post-CCVD impurities have a significant effect on the oxidation process.

Interestingly, the washing process did not seem to alter G/D ratios of non-functionalized, 0 h samples, which indicates that post-CCVD impurities resulting directly from the carbon nanotube synthesis are only present in small quantities. Nevertheless, their amount is enough to alter surface properties as will be evidenced by nitrogen adsorption measurements later.

As discussed previously, the steadily decreasing G/D ratio in the CNT-NW series could probably be attributed to the formation of both oxidation debris and functional groups. On the other hand, washing the nanotubes after oxidation alters G/D ratios considerably, as evidenced by the CNT-AW samples. The G/D ratio of CNT-AW samples exhibits a sudden drop at 4 h of oxidation followed by insignificantly small further changes up until 16 h. This phenomenon must be caused by the washing process: oxidation debris is removed regardless of the functionalization time, leaving only functional groups unaffected by washing on the surface. These groups do not seem alter the G/D ratio of the nanotubes.

G/D intensity ratios of CNT-AW samples are slightly higher in the 4–20 h functionalization time range than those of the corresponding CNT-BW series. This is because the samples of the CNT-AW series were washed only after the oxidation process, thus post-CCVD impurities were still present on the surface of nanotubes during the acidic treatment. Since the graphitic structure of nanotubes is less prone to oxidation then the structurally amorphous impurities, it can be assumed that these post-CCVD impurities are preferentially consumed by the nitric acid, thus exhibiting a shielding effect for nanotube walls. This resulted in a slightly smaller number of defective sites on the walls of the carbon nanotubes compared to CNT-BW samples, where no impurities were present at the time of oxidation and the nitric acid attack could be fully directed at the CNT walls.

Samples with oxidation times higher than 20 h show virtually identical G/D ratios, which could imply that at this stage the concurring defect-generating and defect-consuming processes have reached equilibrium. These results thus indicate, that while oxidation-related debris greatly affects the detected Raman signal, post-CCVD impurities and anchored functional groups barely alter the obtained spectra even though they can have major impact on other attributes of the samples.

3.2 Nitrogen gas adsorption

Nitrogen gas adsorption isotherms were measured for all carbon nanotube samples. The specific surface area (S_BET) was calculated using the well-known Brunauer–Emmett–Teller (BET) method. By inspecting S_BET data plotted in Fig. 3, clear and significant effects of the functionalization time and the acetone washing process are visible.

Fig. 3 Specific surface area of carbon nanotubes subjected to different washing processes as a function of the oxidation time. Lines are guides for the eye only.

In case of CNT-BW samples, S_BET value rises from 225 m² g⁻¹ to 257 m² g⁻¹ at 4 h of oxidation, then it shows a steady drop up to 24 h functionalization where it reaches 178 m² g⁻¹. The initial S_BET value of the CNT-NW sample is 140 m² g⁻¹, then it increases to 205 m² g⁻¹ after 8 h functionalization and decreases again to 155 m² g⁻¹ by 16 h. There is no significant change in specific surface area after 16 h of functionalization for CNT-NW samples. CNT-AW samples do not display the tendencies showed by the CNT-BW and CNT-NW series. Rather, their S_BET of 225 m² g⁻¹ drops to 200 m² g⁻¹ at 4 h and remains fairly constant at this value at longer functionalization times.

The sudden rise of S_BET value on samples CNT-BW and CNT-NW can be attributed to the opening of nanotube endings by the oxidation of the caps in the first few hours of oxidation. The higher C–C bonding curvature and pentagon concentration make these sites more prone to oxidation.³³ The shortening of nanotubes during oxidation also contributes to making the inner pores available, as discussed in the Introduction part. The following steady drop in the specific surface area when using higher functionalization times can be assigned to the formation of oxidation debris composed mainly of peeled off oxidized graphene “islands” forming amorphous carbon³⁴ that can make adsorption sites located in interstitial and intratube pores partly inaccessible for nitrogen.³⁵

Since CNT-AW samples are the purified versions of CNT-NW samples, the measured S_BET values should be directly compared to CNT-NW samples and not inspected completely separately. As a result of the acetone washing, all CNT-AW samples with the exception of 8 h sample show increased specific surface area values if compared to CNT-NW samples, which means that the washing process removed adsorption site blocking impurities from the surface. The change in S_BET was most pronounced at initial non-oxidized samples, where the washing process resulted in an almost 40% increase from 140 m² g⁻¹ to 225 m² g⁻¹, which is evidence of the removal of post-CCVD impurities. It is important to note here, that while washing did not significantly alter the Raman spectra, it did cause a significant change in the specific surface area of non-functionalized samples. This supports the assumption that post-CCVD contamination can greatly alter the surface properties of carbon nanotubes.

At 4 h of oxidation time the difference in S_BET is fairly small between the two series, which means that the acid treatment itself removed most post-CCVD impurities at this stage. The identical values of both series found at 8 h oxidation time suggests that no considerable amount of adsorption site blocking species were left on the surface of nanotubes for the acetone to remove. For functionalization times longer than 8 h the difference in S_BET gradually increases, which suggests that the amount of removed adsorption site blocking species is also gradually getting higher. This phenomenon can be attributed to the newly forming oxidation debris, which block adsorption sites located in interstitial and intratube pores, as disussed previously. Interestingly, the removal of different impurities resulted in a fairly constant specific surface area value of ∼200 m² g⁻¹ for the entire CNT-AW series, which means that the oxidative treatment does not seriously alter the specific surface area of the nanotubes themselves, not even after prolonged exposure. Rather, differences in the S_BET values are related to the blockage of various pores and adsorption sites by the debris formed during the oxidation process. Results for non-oxidized washed samples also indicate that post-CCVD impurities greatly alter S_BET values even in small quantities, even though they are not detectable by Raman spectroscopy.

3.3 Transmission electron microscopy

The structure of carbon nanotubes and supported Pd nanoparticles was examined by transmission electron microscopy (TEM). TEM micrographs of carbon nanotube supported Pd nanoparticles can be found in Fig. 4 and S1.† All particles are evenly dispersed on the surface of the carbon nanotubes. Non-supported particles are not observable. Diameters of the Pd nanoparticles were manually determined for each sample by measuring at least 200 particles per sample. Calculated mean diameters and standard deviations are tabulated in Table 1.

Fig. 4 TEM micrographs of Pd/CNT samples from the “Before Washed” series with nanotube supports oxidized for 0, 12 and 24 h. Numbered arrows highlight distinct features of oxidized samples, such as surface coverage (1 and 2), loose layers of CNT (3), short nanotube pieces (4), CNT fragments (5) and amorphous content (6). The surface of “After Washed” nanotube sample oxidized for 24 h contains no debris or other contaminants.

Table 1 Average diameter and standard deviation of supported Pd nanoparticles

Oxidation time (h)	Particle diameter (nm)
	Pd-CNT-BW	Pd-CNT-NW	Pd-CNT-AW
0	2.7 ± 0.8	4.8 ± 2.3	2.7 ± 0.8
4	2.3 ± 0.4	3.6 ± 1.2	2.7 ± 0.6
8	2.3 ± 0.4	2.9 ± 0.9	2.7 ± 0.7
12	2.2 ± 0.4	2.3 ± 0.5	2.8 ± 0.9
16	2.3 ± 0.5	2.2 ± 0.5	2.9 ± 0.9
20	2.3 ± 0.7	2.3 ± 0.6	2.8 ± 0.6
24	2.6 ± 0.8	2.2 ± 0.5	2.9 ± 0.6

---

The changes in nanotube diameters and length as a function of oxidation time were also determined by manual scaling, and results are depicted in Fig. S6 and S7.† These results also confirm the shortening of nanotubes due to oxidation as discussed in the Introduction section, but interestingly nanotube diameters were found to increase as oxidation time was increased. This phenomenon is due to the acid attacking smaller diameter nanotubes first, as the higher curvature of the graphitic sheet results in higher reactivity.

The formation of metallic Pd was also confirmed by powder XRD measurements, where peaks characteristic for face-centered cubic Pd (111) and Pd (200) reflections were detected on all Pd modified samples at d-spacing values of 2.24 and 1.93 Å, respectively. Calculation of crystallite sizes with Scherrer's equation was not possible due to the pronounced peak broadening typical in this particle size regime. XRD patterns are presented on Fig. S2.†

The average Pd particle diameter was 4.8 nm for Pd-CNT-NW samples without functionalization. This value decreased to 3.6 nm at 4 h, 2.9 nm at 8 h and reached 2.3 nm at 12 h, then remained fairly constant up to 24 h functionalization time. The Pd-CNT-BW and Pd-CNT-AW samples featured an average particle diameter of 2.4 nm and 2.8 nm, respectively independent of the functionalization duration. The standard deviation of the particle diameter was below 1 nm for most of the samples. However, Pd nanoparticles supported on non-functionalized carbon nanotubes showed slightly higher standard deviation of 0.8–2.3 nm (Fig. S4).†

As evidenced by the micrograph of the acetone-washed, non-oxidized (0 h) CNT sample (Fig. 4), the surface of the nanotubes is completely intact. No contamination of any sort is visible on the surface. However, after 12 h of oxidation evidences of surface coverage (arrows 1 and 2, Fig. 4) and some loose layers (arrow 3, Fig. 4) become visible, which are most probably the oxidation debris formed during functionalization. After 24 h of oxidation the results of the extensive acid treatment: amorphous carbon content (arrow 6, Fig. 4) and CNT fragments (arrow 5, Fig. 4) appear on the surface of the carbon nanotubes. Moreover, nanotubes could break up into shorter pieces during prolonged oxidation as evidenced by the Pd-CNT-BW-24 h sample (arrow 4, Fig. 4). No surface contamination can be identified on the Pd-CNT-AW-24 h sample, which provides independent verification for the successful removal of oxidation debris by the Soxhlet-extractor washing.

In the case of CNT-NW samples, oxidation times yielded Pd nanoparticle diameters similar to our earlier observations.¹⁶ The size of the Pd nanoparticles supported on CNT-BW and CNT-AW series of carbon nanotubes seem to be entirely insensitive to the functionalization time. It must be noted however, that in case of washed and not washed unfunctionalized samples the mean diameter changed to 2.7 ± 0.8 nm from 4.8 ± 2.3 nm, respectively. This indicates a major change in the carbon nanotube surface environment due the removal of post-CCVD impurities during the Soxhlet-extractor washing process. Some difference in average particle diameter and standard deviation is still observable between not washed and washed samples oxidized for 4 h. This could mean that 4 h of oxidation may not remove all post-CCVD impurities.

Although the changes in Raman G/D intensity ratios were very similar between the CNT-BW and CNT-AW series, the corresponding S_BET values exhibit considerable differences. Let us now clarify that the dissimilarity revealed by the nitrogen sorption studies does not contradict the previously found Raman similarities between the washed samples, rather, it can help us understand how the washing process alters the environment of the carbon nanotube surfaces and which types of carbon species contribute to Raman spectra or alter the size of Pd nanoparticles with functionalization time. Nitrogen sorption data of CNT-BW samples indicate that oxidation debris is formed during extended functionalization, which lowers S_BET after 4 h. Post-CCVD impurities influence the nanoparticle size and size distribution as evidenced by the TEM studies of the unfunctionalized samples. The presence of oxidation debris does not seriously alter the size of supported particles. This may be attributed to the fact that oxidation debris does not encompass the available surface of the carbon nanotubes, rather, it accumulates in certain spots like interstitial and intratube pore entrances (nanotube endings). Since nanotubes are impregnated with Pd nanoparticles only after functionalization, the inner surfaces of the CNTs are available for Pd nanoparticle anchoring during impregnation from the toluene solution of the Pd(OAc)₂. Oxidation debris does not enter the inner pores of nanotubes, therefore, Pd nanoparticles anchored on the inner walls of nanotubes are simply not affected by the debris. Nanoparticles anchored on the outer walls of nanotubes seem to be equally insensitive to the presence of oxidation debris. These findings suggest that while post-CCVD impurities affect supported Pd nanoparticle size, oxidation debris does not have significant effect on the nanoparticles.

3.4 Cyclohexene hydrogenation

Continuous flow catalytic reaction tests were performed on all samples which are only intended to reveal further details on the effects of functionalization and washing. The hydrogenation of cyclohexene to cyclohexane was monitored and turnover frequencies (TOF) were calculated for all samples after 10 min of reaction time (Fig. 5).

Fig. 5 TOF values of the different Pd decorated CNT catalysts in cyclohexene hydrogenation reaction.

In case of CNT-NW samples, a TOF value of 1.7 molecules per site per s was measured for the unfunctionalized sample, then increased steadily to 2.4 at 4 h and reached 3.9 at 8 h oxidation. No significant changes were observed in the TOF at longer oxidation times. Unfunctionalized Pd-CNT-BW and Pd-CNT-AW samples on the other hand showed no significant cyclohexene conversion, whereas longer oxidation times resulted in TOF values comparable or even larger than those of the corresponding Pd-CNT-NW samples. This can be attributed to the increasing amount of polar, oxygen containing functional groups on the surface with the progression of functionalization. On the other hand, carboxylic groups of the surface enhancing the number of the Brønsted acid sites. These sites are favourable for carbon deposition poisoning of the catalysts as well as may have other effect on the activity change in case of the oxidized samples.³⁶ Interestingly, acetone-washed, non-oxidized (0 h) samples showed almost negligible TOF values compared to the 1.7 molecules per site per s value of the non-washed one, which seems to indicate that the presence of post-CCVD impurities has a significant effect on the catalytic activity. On the other hand, CNT-AW and CNT-BW samples oxidized for 4 h exhibited significant hydrogenation activity (1.65 and 4.8 molecules per site per s, respectively) compared to non-oxidized samples. These results show that the presence of the post-CCVD impurities can inhibit the oxidation process and/or the removal of the oxidation debris can enhance catalytic activity.

Nanotubes oxidized for 8 h or longer show almost identical TOF values for all samples, which suggests that oxidation debris and functional groups do not alter the cyclohexene conversion in these samples as the size of the Pd nanoparticles has not changed with oxidation time in the 8–24 h range. This could be explained by the fact that carbon nanotubes break up into shorter nanotubes during prolonged oxidation, thus the shortening of nanotubes reduces transport resistance making the effect of oxidation debris less significant. Moreover, the inner nanotube surfaces are less prone to get covered by oxidation debris because of the narrower entrance diameter of inner nanotube pores compared to intertubular pores of forming CNT agglomerates. This is important, as it is possible for nanoparticles to be anchored in the interior of CNT and therefore, they are less affected by debris. The effects of nanoparticles residing on the internal side of nanotube walls was also studied, where such catalysts showed better performance and selectivity in various hydrogenation reactions^37–39 due to the reduced reaction volume which probably helps reactants collide more frequently as explained by the collision theory of chemical reactions.^40–42 These phenomena can explain why there is no decrease in TOF for CNT-NW and CNT-BW samples despite the fact that the specific surface area tends to decrease for functionalization times longer than 8 h. Consequently, oxidation debris blocks mostly intertube pores in the forming agglomerates, but not necessarily the inner pores of tubes.

Interestingly, despite the fact that cyclohexene hydrogenation can be considered a structure sensitive reaction,⁴³ no significant differences in TOF were found between the washed and non-washed sample series, even though some kind of difference could be expected due to nanoparticle size variation in the non-washed series. This suggests that surface diffusion and sorption properties of reactants are the main governing factors of reaction rates under such reaction conditions.

4. Conclusions

A new method was devised to study the effects of post-CCVD impurities, functional groups and oxidation debris separately on various properties of carbon nanotubes by incorporating a Soxhlet-extractor enhanced acetone washing process into the standard synthesis method of oxidized carbon nanotubes. “After washed” samples had only functional groups on the surface, while “before washed” samples both oxidation debris and functional groups and finally “not washed” contained all types of surface moieties. This differentiation allowed the determination of the effect of each moiety by the process of elimination, which allowed making the following conclusions:

Post-CCVD impurities were found to reduce the specific surface area severely from 225 m² g⁻¹ to 140 m² g⁻¹ in non-oxidized nanotubes, which suggests a negative effect on specific surface area. It showed a positive effect on cyclohexene hydrogenation TOF on the other hand, as non-washed samples exhibited a value of 2.7 molecules per site per s compared to the value of ∼0.1 molecules per site per s of acetone washed samples. Post-CCVD impurities also affected supported Pd nanoparticle size heavily, but had little effect on Raman G/D intensity ratios.

Oxidation debris was also found to significantly alter the S_BET of carbon nanotubes by blocking intertubular pores, but had little effect on TOF. This phenomenon can be attributed to the shortening of carbon nanotube by prolonged oxidation, which resulted in reduced transport resistance. The presence of oxidation debris has no effect on the size of the supported Pd nanoparticles, but affects Raman G/D intensity ratios.

Functional groups have insignificant effect on the S_BET or Raman G/D ratios. Post-CCVD impurities heavily affect S_BET but do not change G/D ratios. Unlike post-CCVD impurities, functional groups have no effect on the size of supported Pd nanoparticles.

The above results are summarized in a simple graphical representation in Fig. 6, where the effect of individual surface modifying species (post-CCVD impurities, oxidation debris and functional groups) is visually highlighted for various investigated attributes.

Fig. 6 Simple graphical representation of the connections between various attributes of samples and different surface structures (post-CCVD impurities, oxidation debris and functional groups). Tinted areas represent a known effect on the specific attribute by the appropriate surface modifying species, whereas white areas represent a negligible or no effect on an attribute.

Acknowledgements

This paper was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences of SA. The financial support of the OTKA projects NN 110676 and K 112531 is acknowledged.

References

1. C. H. Li, Z. Yu, K. F. Yao, S. F. Ji and J. Liang, J. Mol. Catal. A: Chem., 2005, 226, 101–105.
2. Z. J. Liu, Z. Xu, Z. Y. Yuan, D. Lu, W. Chen and W. Zhou, Catal. Lett., 2001, 72, 203–206.
3. Y. Wang, N. Shah, F. E. Huggins and G. P. Huffman, Energy Fuels, 2006, 20, 2612–2615.
4. G. Wu and B. Q. Xu, J. Power Sources, 2007, 174, 148–158.
5. X. Yang, W. Wang and J. Qiu, Appl. Catal., A, 2010, 382, 131–137.
6. L. P. Biro, N. Q. Khanh, Z. Vertesy, Z. E. Horvath, Z. Osvath, A. Koos, J. Gyulai, A. Kocsonya, Z. Konya, X. B. Zhang, G. Van Tendeloo, A. Fonseca and J. B. Nagy, Mater. Sci. Eng., C, 2002, 19, 9–13.
7. K. Pyrzynska and M. Bystrzejewski, Colloids Surf., A, 2010, 362, 102–109.
8. A. Ricca and C. W. Bauschlicher, Chem. Phys., 2006, 324, 455–458.
9. A. Wongkoblap, D. D. Do and K. Wang, J. Colloid Interface Sci., 2009, 331, 65–76.
10. Z. K. Horastani, S. J. Hashemifar, S. M. Sayedi and M. H. Sheikhi, Int. J. Hydrogen Energy, 2013, 38, 13680–13686.
11. R. Puskas, Á. Kukovecz and Z. Konya, Adsorption, 2013, 19, 501–508.
12. M. Watanabe, T. Akimoto and E. Kondoh, Phys. Status Solidi A, 2012, 209, 2514–2520.
13. K. J. Ziegler, Z. Gu, H. Peng, E. L. Flor, R. H. Hauge and R. E. Smalley, J. Am. Chem. Soc., 2005, 127, 1541–1547.
14. M. W. Marshall, S. Popa-Nita and J. G. Shapter, Carbon, 2006, 44, 1137–1141.
15. G. A. Forrest and A. J. Alexander, J. Phys. Chem. C, 2007, 111, 10792–10798.
16. N. Yao, V. Lordi, S. X. C. Ma, E. Dujardin, A. Krishnan, M. M. J. Treacy and T. W. Ebbesen, J. Mater. Res., 1998, 13, 2432–2437.
17. I. D. Rosca, F. Watari, M. Uo and T. Akasaka, Carbon, 2005, 43, 3124–3131.
18. V. Datsyuk, M. Kalyva, K. Papagelis, J. Parthenios, D. Tasis, A. Siokou, I. Kallitsis and C. Galiotis, Carbon, 2008, 46, 833–840.
19. F. Aviles, J. V. Cauich-Rodrıguez, L. Moo-Tah, A. May-Pat and R. Vargas-Coronado, Carbon, 2009, 47, 2970–2975.
20. S. Kundu, Y. Wang, W. Xia and M. Muhler, J. Phys. Chem. C, 2008, 112, 16869–16878.
21. J. Zhang, H. Zou, Q. Qing, Y. Yang, Q. Li, Z. Liu, X. Guo and Z. Du, J. Phys. Chem. B, 2003, 107, 3712–3718.
22. R. Verdejo, S. Lamoriniere, B. Cottam, A. Bismarck and M. Shaffer, Chem. Commun., 2007, 5, 513–515.
23. S. Fogden, R. Verdejo, B. Cottam and M. Shaffer, Phys. Lett., 2008, 460, 162–167.
24. D. Stefani, A. J. Paula, B. G. Vaz, R. A. Silva, N. F. Andrade and G. Z. Justo, J. Hazard. Mater., 2011, 189, 391–396.
25. Z. Wu, R. F. Hamilton, Z. Wang, A. Holian and S. Mitra, Carbon, 2014, 68, 678–686.
26. Z. Wu and S. Mitra, Carbon, 2015, 88, 233–238.
27. N. Mahata, K. V. Raghavan and V. Vishwanathan, Appl. Catal., A, 1999, 182, 183–187.
28. S. Tracey, A. Palermo, J. P. H. Vazquez and R. M. Lambert, J. Catal., 1998, 179, 231–240.
29. Y. H. Park and G. L. Price, Ind. Eng. Chem. Res., 1992, 31, 469–474.
30. A. Kukovecz, Z. Konya, N. Nagaraju, I. Willems, A. Tamasi, A. Fonseca, J. B. Nagy and I. Kiricsi, Phys. Chem. Chem. Phys., 2000, 2, 3071–3076.
31. P. K. Gallagher and M. E. Gross, J. Therm. Anal., 1986, 31, 1231–1241.
32. C. J. Lee, J. Park, Y. Huh and J. Y. Lee, Chem. Phys. Lett., 2001, 343, 33–38.
33. P. M. Ajayan, T. W. Ebbesen, T. Ichihashi, S. Iijima, K. Tanigaki and H. Hiura, Nature, 1993, 362, 522–525.
34. E. Dujardin, T. W. Ebbesen, A. Krishnan and M. M. Treacy, Adv. Mater., 1998, 10, 611–613.
35. S. Agnihotri, J. P. B. Mota, M. Rostam-Abadi and M. J. Rood, Carbon, 2006, 44, 2376–2383.
36. R. Puskas, A. Sapi, A. Kukovecz and Z. Konya, Top. Catal., 2012, 55, 865–872.
37. H. Ma, L. Wang, L. Chen, C. Dong, W. Yu, T. Huang and Y. Qian, Catal. Commun., 2007, 8, 452–456.
38. E. Castillejos, P. J. Debouttiere, L. Roiban, A. Solhy, V. Martinez, Y. Kihn, O. Ersen, K. Philippot, B. Chaudret and P. Serp, Angew. Chem., 2009, 121, 2567–2571.
39. A. M. Zhang, J. L. Dong, Q. H. Xu, H. K. Rhee and X. L. Li, Catal. Today, 2004, 93–95, 347–352.
40. T. Lu and E. M. Goldfield, J. Phys. Chem. C, 2008, 112, 2654–2659.
41. E. E. Santiso, M. K. Kostov, A. M. George, M. Buongiorno-Nardelli and K. E. Gubbins, Appl. Surf. Sci., 2007, 253, 5570–5579.
42. M. D. Halls and H. B. Schlegel, J. Phys. Chem. B, 2002, 106, 1921–1925.
43. V. V. Pushkarev, K. An, S. Alayoglu, S. Beaumont and G. A. Somorjai, J. Catal., 2012, 292, 64–72.

---

Footnote

† Electronic supplementary information (ESI) available: TEM, XRD, Raman spectra, particle diameter and calculations. See DOI: 10.1039/c6ra16918j

---