Carbon Nanotubes Decorated with Platinum Nanoparticles for Field-Emission Application.

In this work, a simple and scalable method was presented to decorate carbon nanotubes (CNTs) with platinum nanoparticles (Pt NPs) by atomic layer deposition (ALD) for field emission (FE) application. The size and distribution of NPs were precisely controlled by adjusting the number of ALD cycles. It was discovered that a higher cycle number of ALD resulted in continuously increased conductivity of the CNTs@Pt nanocomposite. It could be explained by not only the increased loading of Pt NPs but also by the gradually decreased presence of the PtO compound in NPs. As a result, a significant improvement in the FE characteristics of the cold cathode was observed because of the decoration of CNTs with metal NPs. The higher number of ALD cycles resulted in the gradual lowering of the turn-on electric field which reached a minimum value of ∼0.43 V/μm after 100 cycles of ALD.

Introduction

Since their discovery in 1991 by Iijima, carbon nanotubes (CNTs) have attracted great interest in most fields of science and engineering given their unique physical and chemical properties.[1−3] For instance, due to the low electron affinity and relatively low effective work function in addition to the high electrical and thermal conductivities, mechanical strength, excellent chemical stability, and exceptionally high aspect ratio, CNTs have been considered as one of the most promising candidates for a cold cathode field emitter.[4]

In recent years, several approaches to improve field emission (FE) properties of CNTs and to make them suit the present needs were reported. Among them doping, coating, thermal annealing, cutting, and tip treatment of CNTs deserved special attention as it proved to be an efficient tool to not only greatly decrease its turn-on electric field but also to extend its stability for tens of hours.[5] For instance, Sun et al. reported that annealing at 900 °C could influence significantly on FE properties of CNTs.[4] In turn, the surface modification of CNTs by decorating with metal nanoparticles (NPs) was discovered to be one of the most effective methods to enhance FE performance.[6] In particular, it was reported that the modification of the electronic structure of CNTs by decoration of its surface with Co, Ti, Pd, W, and Ru NPs resulted in highly increased density of states near the Fermi level, which was caused by charge transfer and orbital hybridization.[6,7] As a result, the FE properties of such composite structures improved significantly compared with that of pristine CNTs. Yet, it is still unclear how the size and chemical composition of metal NPs can influence and control the FE characteristics of CNTs, as most of the published reports concentrated on searching for new materials or compounds to decorate the surface of CNTs, while a detailed experimental investigation of their structural particularities and its correlation with improved performance of CNTs was covered insufficiently.[8,9] Another important issue which needs to be raised is that to fully and uniformly decorate CNTs with metal NPs at a size of several nanometers given to a dense forest structure, high-aspect ratio morphology, and profound layered thickness of CNTs can be a nontrivial and challenging task. Often the reported methods, such as, for example, the magnetron sputtering[6] or chemical reduction method[7] might suffer from low conformability or inability to precisely control the size of NPs down to several nanometers.

To solve this problem, in this work, we demonstrated the employment of the atomic layer deposition (ALD) technique to prepare CNTs decorated with sized- and composition-controlled Pt NPs. Because of its sequential, self-limiting surface reactions, the ALD brings various advantages over other alternative deposition techniques such as, for example, chemical vapor deposition (CVD) and physical vapor deposition, not to mention others.[10] In fact, ALD provides excellent step coverage and unprecedented conformality on high-aspect ratio structures, control over thickness at an atomic-scale level, and tunable film composition and stoichiometry.[11] The platinum as model metal NPs to deposit on the surface of CNTs was chosen due to its great potential in related fields such as fuel cells and electrocatalysts.[12] Furthermore, platinum recommended itself as an efficient material to enhance the FE characteristics of other materials such as, for example, SiO2 nanowires.[13] Overall, it is believed that the CNTs@Pt composite might have great potential for practical application as an efficient field emitter.

Experimental Section

The catalyst for the growth of CNTs was prepared by mixing iron nitrate, tetrabutyl titanate, and n-propanol. More details can be found in our previous work.[14] The solutions were vigorously stirred for 1 h and later sonicated for 30 min to completely dissolve the iron nitrate. An n-type silicon wafer (100) was used as a substrate material for the growth of CNTs. The substrate was first rinsed in acetone and later was blown with clean, dry, and compressed nitrogen. In the following step, the catalyst was spin-coated on the substrate at 6000 rpm for 30 s. The coated wafers were quickly placed on the Mo substrate upside down and transferred into home-made microwave plasma-enhanced CVD (MPCVD). Prior to deposition, the reaction chamber was evacuated to ∼3 × 10–2 Torr by a mechanical pump. CNTs were deposited with a microwave power of 800 W at 40 Torr using the mixture of CH4/H2/N which flow rates were controlled at 20, 40, and 50 sccm, respectively. The temperature inside the reaction chamber was raised naturally by plasma ignition. It reached 449 °C in the end of deposition. Once the sample with grown carbon nanostructures was naturally cooled down, it was immediately subjected to the oxygen plasma treatment, which was held at a power of 600 W and flow rate of 20 sccm. As it is not easy to trigger the attachment of Pt NPs on pristine CNTs by ALD due to their inert surface, to apply the postsynthesis oxygen plasma treatment is a necessary step to tailor the surface chemical composition of CNTs.[15] After several attempts, the duration of plasma treatment was chosen ∼5 s, as longer time seriously affects the morphology of carbon nanostructures, while shorter time results in insufficient density of functional groups. Next, the sample was directly transferred to the home-made ALD (Syskey Technology) for the synthesis of Pt NPs. To complete this process, (methylcyclopentadienyl)trimethylplatinum (MeCpPtMe3) and oxygen were used as the precursors. The Pt precursor was held in a stainless-steel bubbler at 46 °C, and oxygen was maintained at room temperature. The durations of pulse time of MeCpPtMe3 and O2 were 1.5 and 5.0 s, respectively. After each pulse of the precursor, a purge with nitrogen for 15 s was performed. The ALD of Pt was conducted at 280 °C and the total numbers of cycles were 25, 50, and 100 cycles. Following this, after deposition, the samples were named as CNTs@Pt25, CNTs@Pt50, and CNTs@Pt100, respectively. The fabrication procedure schematic is shown in Figure .

Figure 1

Schematic illustration of the process for fabrication of the CNTs@Pt nanocomposite.

The morphologies and the microstructures of the samples were characterized by the FE scanning electron microscopy (FESEM, Hitachi SU-8010) and transmission electron microscopy (TEM, JEOL JEM-ARM200FTH). The size of Pt NPs was evaluated statistically by measuring a diameter of 160 random NPs for each designated cycle number. The phases and crystal structures were analyzed by grazing incidence X-ray diffraction (TTRAX III, Rigaku) using Cu Kα radiation and an incident angle of 0.5°. The high resolution X-ray photoelectron spectroscopy (HRXPS) analysis of the samples was performed using a PHI Quantera AES 650 X-ray photoelectron spectrometer, which was equipped with a hemispherical energy analyzer. The residual gas pressure in a chamber during data acquisition was lower than 1 × 10–8 Torr. The spectra were measured at room temperature using a monochromatic Al Kα X-ray source (25.5 W, spot size is 100 micrometer), a take-off angle of 45°, and pass energy of 280 eV. A conductive atomic force microscope (Bruker, Dimension ICON, scan rate 1 Hz, resonant frequency 1.3 MHz) attached with peak-force tunneling mode (PF-TUNA) was used at an applied bias of 0.8 V to probe the electrical currents of CNTs before and after coating with Pt NPs. The utilization of this mode requires the use of a substrate with enough conductivity to pass the current to the sample while it also should not be too conductive because complete saturation of the current might occur.[16,17] Thus, it is justified the choice of Si wafer as a substrate for measurement of CNTs electrical properties.

The FE characteristics of the materials were measured using Keithley 2410, 2290, and 6485 systems in a high-vacuum chamber under a pressure of 3 × 10–6 Torr. For the measuring of the FE characteristics, the distance between the anode and cathode was kept at 300 μm and the area of the tungsten anode was 0.03 cm2.

Results and Discussion

X-ray photoelectron spectroscopy (XPS) analysis was performed to determine the chemical modification of CNTs before and after oxygen plasma treatment. The XPS survey scan spectra for both CNTs are shown in Figure a. The peak located at 284.5 eV in all spectra is due to the photoelectrons emitted from the C 1s core level, and the peak at 535.0 eV is generated by those from the O 1s core level. The concentrations of carbon and oxygen calculated from spectra were 91.8 and 8.2% versus 88.4 and 11.6% for CNTs before and after plasma treatment, respectively. It is seen that the concentration of oxygen increases after plasma treatment while that of carbon decreases at the same time. This suggests that oxygen is successfully grafted onto the CNT surface, which is in accordance with a previous report.[15]Figure b,c shows the HRXPS C 1s spectra of CNTs before and after plasma treatment. In order to further explain the process of plasma oxidation, the C 1s peak was deconvoluted into three component Gaussian peaks, a C–C bond at 284.6 eV, a C–O bond at 285.4 eV, and an O–C=O bond at 287.7 eV.[18] The relative percentages of these carbon atoms before and after plasma treatment are listed in Table . The concentration of C–C bonds in pristine CNTs is greater than that in treated CNTs, while in turn the treated CNTs have higher percentages of oxygen-containing functional groups after exposure to plasma. This implies that the treated CNTs may have more defects, which can act as nucleation sites for the growth of Pt NPs.

Figure 2

(a) XPS survey spectra of CNTs before and after oxygen plasma treatment. XPS C 1s spectra of (b) pristine CNTs and (c) plasma-treated CNTs. The curve fitting suggests the existence of three species.

Table 1

Relative Percentages of the Three Component Peaks of the Carbon Atoms before and after Oxygen Plasma Treatment for  5 s

sample	C–C	C–O	O–C=O
pristine CNTs	67.3	13.7	19.0
plasma-treated CNTs	42.7	33.9	23.4

The morphology of CNTs before and after oxygen plasma treatment was investigated by FESEM. As can be seen from Figure a, the as-grown NTs represent curly and twisted structures with some agglomerations. The length of NTs was determined to be ∼10–12 μm in Figure a. According to our previous work, the grown NTs are attributed to the multiwall nanostructures.[14] As for the sample subjected to oxygen plasma treatment, its morphology remained the same as that of pristine sample (Figure c,d). Figure e–g represent CNT samples decorated with Pt NPs fabricated by ALD with 25, 50, and 100 cycles, respectively. Low magnification SEM images revealed that the morphology of CNTs was not influenced much by the ALD process even though it was performed at a significantly high temperature (∼280 °C). Pt NPs deposited with 25 cycles are randomly spread on the surface of CNTs with an irregular size (inset in Figure e). Yet, when the Pt NPs were deposited with 50 cycles, their density and distribution improved significantly (insets in Figure f). After deposition for 100 cycles (Figure g), the particle density reached maximum, which resulted in their apparent transformation into thin films (inset in Figure g). The compositional analysis by energy-dispersive X-ray spectrometry (EDX) (Table ) confirms the deposition of Pt NPs on the surface of CNTs and its increased loading with a higher cycle number of ALD.

Figure 3

FESEM images of the surface morphology of CNTs and CNTs@Pt nanocomposites. (a,b) Pristine CNTs (the inset shows the cross section image of the CNT film), (c,d) CNTs after oxygen plasma treatment for 10 s, (e) CNTs@Pt25, (f) CNTs@Pt50, and (g) CNTs@Pt100 [the insets in (e–g) shows higher magnification images].

Table 2

Chemical Compositions (at. %) of CNTs before and after Decoration with Pt NPs by EDX Analysis

sample	C	Pt
pristine CNTs	100	0
CNTs@Pt25	99.94	0.06
CNTs@Pt50	99.42	0.58
CNTs@Pt100	99.26	0.74

In order to precisely determine the size of Pt NPs, TEM and high-resolution TEM (HRTEM) images of CNTs covered with 25, 50, and 100 cycles of Pt ALD were performed. As can be seen from Figure a, CNTs@Pt25 demonstrated the nonuniform distribution of Pt NPs which size distribution was evaluated statistically to be ∼2.08 ± 1.34 nm (Figure b). It is seen that a certain portion of particles with sizes over 4–5 nm can be clearly visible in the TEM image (Figure a), and their contribution into the overall platinum mass lies in NPs is considered to be sufficient. Such a wide size distribution of Pt NPs deposited with a low cycle number (∼25 cycles) was reported previously and the following explanation was provided.[19,20] There existed two growth regimes for Pt ALD, one at low temperatures (i.e., T ≤ 100 °C) and one at high temperatures (i.e., T ≥ 150 °C). The high temperature regime is characterized by the formation of large NPs aside of small ones and very broad distribution of the interparticle distances. It was noticed, for example, that “10 cycles of conventional thermal Pt ALD...carried out at 250 °C gave rise...to numerous populations of small NPs of about 1 nm coexisting next to NPs as large as 24 nm”.[20] The origin of large Pt NPs could be attributed to the Ostwald ripening,[21] that is, to minimize the surface energy, the growth of larger NPs occurrs at the expense of smaller ones via exchange of single atoms. Such phenomena become possible due to the formation of the PtO species at the surface of a Pt cluster, which lowers the barrier for Pt atom detachment and enables Ostwald ripening at relatively low temperatures. In addition, PtO species (PtO or PtO2) formed in an oxidizing atmosphere might diffuse with higher rates in comparison to metallic Pt atoms.[21] Sufficient presence of the PtO compound in Pt NPs deposited with 25 cycles is demonstrated below. As an alternative, it was also reported that the presence of large Pt NPs could also be explained by NPs diffusion and coalescence due to the different magnitudes of the local temperature and pressure fluctuations during the deposition process.[19,20]

Figure 4

TEM and HRTEM images of CNTs@Pt25 (a,c), CNTs@Pt50 (d,f), and CNTs@Pt100 (g,i) nanocomposites. Figures (b,e,h) show corresponding histograms of the particles size distribution.

For the HRTEM measurement, the particle with sizes over 5 nm was chosen for analysis as it allows to more precisely determine its lattice spacing. The image shown in Figure c reveals that Pt NPs deposited with 25 cycles have a lattice spacing of 0.22 nm, corresponding to the (111) plane of face centered cubic Pt, which is in good agreement with the XRD pattern (will be shown below).[12] It is also noted that the intershell distance of CNTs clearly seen in Figure c is 0.35 nm and it satisfies the values represented in literature.[22] In turn, the sample, prepared with 50 cycles of Pt ALD shows a denser distribution of Pt NPs with very regular size (Figure d) which could be estimated as ∼2.96 ± 0.48 nm (Figure e). The lattice spacing of Pt NPs after 50 cycles (0.22 nm) was determined to be identical to that of Pt deposited with 25 cycles (Figure f). Finally, the sample prepared with 100 cycles of Pt ALD demonstrates a very high presence of NPs, which are often overlapped with each other (Figure g). The particle size was statistically determined to be ∼5.16 ± 0.82 nm (Figure h), it is clearly seen that the size of NPs increases sufficiently as well as their density. The lattice spacing revealed by the HRTEM analysis (Figure i) is determined to be 0.22 nm, similar to that prepared with 25 and 50 cycles of ALD.

XRD analysis (Figure ) was utilized to investigate the crystal structure of the samples. The XRD pattern taken from the uncoated CNTs reveals a peak at around 27° which is corresponding to hexagonal carbon (002).[23] In addition, several peaks ascribed to Fe3O4 are also could be observed. Their appearance is explained by the decomposition of iron nitrate NPs during the growth of CNTs as the temperature inside the MPCVD chamber reached over 400 °C.[24]

Figure 5

(a) XRD patterns of CNTs before and after decoration with Pt NPs. (b,c) represents enlarged and deconvoluted XRD patterns in the selected range of the pristine CNTs and CNTs@Pt25 samples, respectively.

After the deposition of Pt NPs by ALD, the samples deposited with 50 and 100 cycles of ALD revealed clear appearance of peaks at ∼39.7° and ∼46.3°, which belong to (111) and (200) orientations, respectively, and it is in agreement with that reported by Aaltonen et al.[25] For the sample deposited with 25 cycles of ALD, the observation of peaks ascribed to Pt NPs become possible only after the deconvolution of the XRD pattern and its consequent comparison with pristine CNTs (Figure c). Overall, the gradually increased intensity of Pt peaks follows the ALD cycle number, which is explained by higher loading and improved crystallinity of NPs. As a final remark, it is important to mention that the samples covered with Pt NPs demonstrate the lower intensity of the graphite peak (002) which gradually declines following the cycle number of ALD as the coverage of the CNT surface with Pt NP increases.

XPS was performed to further analyze the composition of Pt NPs and to probe the effect of the Pt NP size on the electronic properties of CNTs@Pt nanocomposites. Figure shows the high-resolution Pt 4f spectra of the CNTs@Pt nanocomposites which exhibit doublets from the spin–orbit splitting of the 4f7/2 and 4f5/2 states and intensities of them are increased with the increased cycle number of ALD. In addition, they can be deconvoluted into two peaks, indicating that all Pt NPs consist of metallic Pt and PtO. The presence of PtO can be explained by the ALD reaction, where the oxygen precursor is involved in the formation of Pt.[26] Moreover, it is also highly possible that the very small size of Pt NPs is susceptible to oxidation by the ambient environment. After careful evaluation of the relative integrated area percentages of the peaks referred to Pt and PtO, it is seen that the content of PtO decreases with the increased cycle number of ALD (Table ). It was reported that at a low cycle number, the “initial formation of PtO”[27] might occur, the presence of which becomes less evident with increased cycle number of ALD. It is in accordance with our observation. Finally, given that the penetration depth of XPS is around 10 nm,[11] it allows to precisely calculating the atomic ratio of Pt on the surface of CNTs. The following values of 0.9, 14.2, and 44.3% are determined for CNTs@Pt25, CNTs@Pt50, and CNTs@Pt100, respectively. As can be seen, with the increased cycle number of ALD, the loading of Pt sufficiently increases which is in agreement with XRD and EDX results.

Figure 6

XPS Pt spectra of (a) CNTs@Pt25, (b) CNTs@Pt50, and (c) CNTs@Pt100 nanocomposites. The curve fitting suggests the existence of two species.

Table 3

Relative Integrated Area Percentages of the Component Peaks of Platinum Atoms for the Samples Prepared with 25, 50, and 100 Cycles of ALD

sample	Pt	PtO
CNTs@Pt25	59.6	40.4
CNTs@Pt50	66.2	33.8
CNTs@Pt100	76.9	23.1

In order to determine the electrical properties of CNTs before and after decoration with Pt NPs by ALD, conductive atomic force microscopy was applied. As can be seen from Figure , the pristine CNTs show the maximum current amplitude at about 1.2 nA. On the other hand, after Pt NPs were deposited with 25, 50, and 100 cycles of ALD, the current amplitude increased to 246.4, 534.7, and 884.7 nA, respectively. Thus, the surface electrical conductivity of the CNTs@Pt nanocomposite is highly increased after the deposition of Pt NPs and the rise of conductivity follows the cycle number of ALD. It could be explained as follows. According to the XPS analysis, the content of PtO in Pt NPs decreases with a higher cycle number of ALD while the presence of metallic, that is, Pt compound increases and become gradually dominated. The conductivity of Pt is obviously higher than that of PtO, thus, decreasing the PtO content would certainly resulted in increased conductivity of Pt NPs.

Figure 7

Two-dimensional maps of peak currents measured by PF-TUNA on (a) pristine CNTs, (b) CNTs@Pt25, (c) CNTs@Pt50, and (d) CNTs@Pt100 nanocomposites.

FE properties of the samples before and after the ALD deposition of Pt NPs  were measured. The cathode was consisted of uncoated CNTs or CNTs@Pt (emitters) grown on Si wafer (substrate) while the anode represented tungsten. The pristine CNTs were chosen as a standard to compare. It is necessary to notice that plasma treatment for ∼5 s has almost no impact on its FE characteristics (Figure S3). Figure a shows current densities at different electric fields of the CNTs with and without the decoration of Pt NPs. The turn-on field for the pristine CNTs, CNTs@Pt25, CNTs@Pt50, and CNTs@Pt100 were calculated to be ∼1.20, ∼0.69, ∼0.57, and ∼0.43 V/μm, respectively. As can be seen, the higher cycle number of ALD causes CNTs to display a lower turn-on field. To further understand the FE characteristic, the J–E data were analyzed using the classical F–N theory, Figure b. The linear curve of the high field segments of the F–N plot indicates the electron emission from the tunnel effect, and the slope of the curve is related to the FE enhancement factor (β). In the F–N theory, β can be calculated using the following equation[14]where B = −6.83 × 103 eV–3/2 V/μm–1, m is the slope of high field segments of F–N plot and ψ is the work function of the emitter material with the assumption of 4.5 eV.[14] The enhancement factor β for the pristine CNTs, CNTs@Pt25, CNTs@Pt50, and CNTs@Pt100 were calculated to be 3571, 6715, 7537, and 17340, respectively.

Figure 8

(a) J–E curves and (b) the corresponding F–N plots for the CNTs before and after decoration with Pt NPs.

Thus, it is obvious that the deposition of Pt NPs positively influences on FE characteristics of CNTs. The higher the cycle number of ALD, the more evident the presence of Pt NPs as their distribution and concentration increased significantly, the higher the conductivity of CNTs@Pt nanocomposites due to the lower presence of the PtO compound, the better FE characteristics. In future, it is possible to extend further the efficiency of the CNTs@Pt composite through more precise and accurate tuning of ALD deposition parameters.

Conclusions

In this work, a simple approach was used to decorate CNTs with Pt NPs by ALD. With different cycle numbers of ALD it was possible to precisely adjust the size and distribution of NPs. The nanocomposite CNTs@Pt showed highly enhanced filed emission characteristics compared with that of pristine CNTs. It was discovered that the higher number of ALD cycles, the greater decrement of turn-on field, which reaches a minimum value of ∼0.43 V/μm after 100 cycles of ALD. The increased performance with increased cycle number of Pt ALD was mostly attributed to the gradually decreased presence of the PtO compound in Pt NPs which resulted in highly enhanced conductivity of the nanocomposite and increased loading.