Preparation of Pt/CNT Catalyst with High Dispersion Structure via Plasma Jet

A plasma jet method based on free arc was carried out to disperse carbon nanotubes (CNTs) in gas phase and produce Pt nanoparticles (Pt-NPs) on them at the same time. The plasma jet is generated through the discharge between the cathode composed of CNTs, PtCl4 and ID water and the anode with a hole. Plasma jet prepares the Pt/CNT catalyst by directly spraying. The CNT carrier in the catalyst have good dispersion by SEM images. The mean size of Pt-NPs estimated by TEM images is 3.41nm with the plasma jet method and about 4.77nm with the commercial Pt/C respectively. It was showed that the bulk of Pt-NPs produced by plasma jet is in metal state in the analysis of XPS measurements. Compared with Pt/C, Pt/CNT have roughly equal catalytic activity and higher durability by cyclic voltammetry measurements. On the basis of these properties, the plasma jet method greatly simpli�es the preparation process of the catalyst by simultaneously solving carrier dispersion and metal particle preparation.


Introduction
CNTs are suitable for supporting precious metal catalysts due to their high speci c surface area, good stability and conductivity, such as Pt/CNT and Pt-Ru/CNT used in fuel cells.The preparation of Pt/CNT catalysts mostly adopts chemical methods, including ordinary liquid phase redox method [1] , solution-gel method [2] , solid phase reaction method, pre-precipitation method [3] , immersion reduction method [4] , etc.
However, CNTs must be in a well-dispersed state before reducing Pt on the surface of CNTs via above methods [5] .In general, the dispersion of multi-walled carbon nanotubes (MWCNTs) is relatively easy.But complex pretreatment [6] , mechanical dispersion [7] and a large amount of dispersant [8] is also required in this process.However, the dispersant must be removed by calcination in the later stage to prevent coating the active particles in the catalyst [9] .
In previous studies, we have found that free arc can effectively disperses CNTs under appropriate conditions [10][11] .This method is applied to lm [12] and composite materials [13] .And good results were achieved.In this work, We continue to explore and nd that while free arc disperses CNTs, it can also reduce Pt ions to form Pt nanoparticles (Pt-NPs) on the surface of CNTs and nally prepare Pt/CNT catalysts.Obviously, Pt ions in various valence states can be easily reduced by reducing agents.
Moreover, many platinum compounds (such as PtCl 4 , H 2 PtCl 6 , etc.) can decompose under high temperatures.In the process of CNTs dispersion, the free arc is used as a strong heat source for the vaporization of the working uid.And it also possesses the high temperature conditions to initiate the decomposition of platinum compounds [14] .
High temperature has a positive effect on the formation of Pt nanoparticles, in which Pt 4+ can be rapidly reduced at high temperature to form Pt 2+ or Pt monomers [15] .Meanwhile, strong heat has a strong evaporation effect on the Pt monomers obtained by chemical reduction, and small metal particles of nanoscale can be easily formed by Physical Vapour Deposition (PVD) mechanism [16] .
With this idea, this paper designs a simple electrode structure to realize a small plasma jet based on the thermal effect of the free arc, achieving the gas phase dispersion of CNTs and the one-step, rapid and high-quality Pt/CNT catalyst preparation at the same time.

Method and equipment
The effect of free arc on the cathode is too rapid to fully reduce Pt ions.For this reason, a special electrode structure as shown in Fig. 1 was designed to extend the heat's action time on the CNTs by generating a plasma jet.To explore the effects of free arc direct action and plasma jet action respectively, the sample prepared with the open electrode (free arc process only) is named CNT-Open and the sample prepared with the two processes (free arc and plasma jet) is named CNT-Jet.
The self-made device is composed of movable and automatically replenished CNTs Cathode and a cooled conical tungsten Anode.There is a 1mm gap between electrodes to generate free arc.And whether the compressed air(0.2MPa)passes through the gap determines plasma jet will be generated or not.The formula of CNT Cathode used in CNT-Open and CNT-Jet is the same: 18.4wt% CNTs, 7.95wt% PtCl 4 , 73.6wt% ID water.All samples are produced in the form of aerogel and collected by negative pressure ltration (except SEM characterization).

Characterization of the Pt/CNT
Field emission scanning electron microscope (SEM) (SU-8010; Hitachi, Tokyo, Japan) was adopted to characterize the morphological distribution of PT/CNT and Pt/C.High resolution transmission electron microscope (HRTEM) (Tecnai G2, 200kV) was adopted to determine the interior microstructure of the CNTs and the morphology and size distribution of the Pt particles that are deposited on the CNTs and C. Thermogravimetric analyzer (TGA) (Dynamic TGA Q500 in TA Instrument 5100) measured the oxidation resistance of the samples, owing air (60cm 3 /min) with a heating rate of 10°C/min.The residual mass is used to estimate the Pt content that is deposited on the surface of the CNTs.
XRD patterns were measured on a D8 ADVANCE diffractometer by using Cu Ka radiation (k = 0.1541nm, 36kV, 2mA, scanning step = 2°/min).The diffraction patterns were recorded by scanning at an angle ranging from 5 to 80°.XPS measurements were conducted by using a PHI Quantera microprobe (ULVAC-PHI Inc., Japan) equipped with an aluminum anode as the monochromatized X-ray source (1486.7eVrun at 10kV and 15mA in xed analyzer transmission mode).The peak tting procedure was performed with XPS Peak 4.1.The C-C peak was set to 284.8eV.

Electrochemical characterization
A CHI 613C electrochemical workstation (CH Instruments) was employed for the electrochemical study of carbon supported Pt samples.A three-electrode electrochemical cell was constructed for CV measurements, through which the ECSA of the Pt nanoparticles was determined.The working electrode was a thin layer of Na on®-impregnated catalyst sample, casting on a vitreous carbon disk of 5 mm in diameter embedded in a Te on cylinder.A Pt wire and a saturated calomel electrode (SCE) were used as the counter and reference electrodes respectively.The measurements of CV were conducted at room temperature, using 0.5M H 2 SO 4 as the electrolyte solution at a scan rate of 20mV/s from − 0.2 to 1.0V vs. SCE.

Results And Discussion
Figure 2a shows the TEM image of the Pt/CNT-Jet.The sizes of the Pt-NPs on the CNTs support was obtained by measuring the diameter of over 500 metal nanoparticles.The prepared catalyst particles are well dispersed with mean size of 3.41nm.While the commercial catalyst particle size is about 4.77nm (Fig. 2b).The CNTs loaded with Pt-NPs directly ejected by the plasma device have a well dispersion state (Fig. 2c), which has a u er structure than commercial catalyst (Fig. 2d).From the perspective of particle size distribution, the particle sizes of Pt/CNT are obviously smaller than that of Pt/C as a whole.But the distribution is wider, especially in the part of large particle sizes.It is also found that individual Pt particles are over 10nm in Pt/CNT samples by TEM.This indicates that the plasma jet mothed is slightly less stable than chemical methods in preparing Pt nanoparticles.
The TGA pro le measured in owing air for all samples are shown in Fig. 3.
The original CNTs show about 2% of ash which is the catalyst component and impurities in the production of CNTs.Although the surface of CNTs is protected by the working uid in the process of dispersing CNTs by free arc, the CNT structure is destroyed to a certain extent under the strong heat of free arc, which decrease the thermal stability of CNTs.The decomposition of CNTs during the dispersion process makes ash increase from 2-3.5%; even so, the CNTs dispersed by free arc still have high thermal stability compared with Pt/C.It is well known that the oxidation of carbon supports was catalyzed by Pt in air atmosphere [17] .Therefore, Pt/CNT-Jet and Pt/CNT-Open shows a lower decomposition temperature compared with dispersed CNT and original CNT.The Pt content of Pt/C is about 21%.After taking into account the in uence of ash, the Pt content of Pt/CNT-Jet and Pt/CNT-Open are respectively 21% and 20.5%, which are basically consistent with the designed 20% content.All of the decomposition temperatures of PtCl x are above 370°C (PtCl 4 :370-435°C, PtCl 3 :435-581°C, PtCl 2 > 581°C).So the quality of Pt/CNT-Open before 370°C degrades because of the crystal water of Pt salt decomposition.This indicates the dispersion of open electrodes does not su ciently decompose PtCl 4 .Pt/CNT-Jet has relatively higher Pt purity.Because the Pt salt on CNT is continuously reduced during the motion in plasma jet.
The deconvolution of the XPS spectra over the Pt4f region (Fig. 4) shows different states of oxidation for each of the carbon supported Pt samples, which consist of three couples of doublets.The most intense doublet, at 71.0 and 74.35eV, represents a zero-valence metallic Pt(Pt 0 ), the doublet at 72.4 and 75.75eV is attributed to the presence of an amorphous Pt(II) species, such as PtCl 2 or Pt(OH) 2 [18]   , and the broader doublet at 74.9 and 78.25eV is assigned to the Pt(IV) species.Table 1 summarizes the calculated percentages of the Pt species in different chemical states.The data shows that the Pt valence distributions of Pt/CNT-Jet and Pt/C are very consistent.Most of the Pt in these two samples is in the zero-valence metallic state (70.7-70.3%).But compared with those of Pt/CNT-Jet and Pt/C, Pt/CNT-Open has the lowest percentage of Pt(Pt 0 ) (46.4%) and the highest percentage of Pt(IV).This fact implies that the direct thermal effect of free arc on the CNTs cathode cannot completely eliminate Pt(IV), the Pt(Pt 0 ) conversion rate in this process is about 65.6% (46.4/70.7×100%).So it can be further seen that the conversion of the remaining 34.4% Pt(Pt 0 ) is completed in the plasma jet process initiated by the free arc.Table 1.Results for the t of the XPS Pt4f region, given as a percentage of the individual total intensity.XRD patterns of both Pt/CNT-Jet and Pt/C catalysts are show in Fig. 5.The diffraction peak at a 2θ value of -26°is attributed to the graphite-like structure, C(002).Peaks at 2θ = 39.7,46.46 and 67.7 are assigned to the (111), ( 200) and (220) planes of Pt respectively, displaying a typical diffraction pattern of fcc lattice for Pt.This indicates that the Pt catalysts have a face-centered cubic structure.The diffraction peaks for Pt/CNT are broader than Pt/C, which shows that the average size of the Pt-NPs on Pt/CNT is smaller than commercial catalyst.The calculated average crystallite size of Pt for Pt/CNT and Pt/C was 6.3 and 3.8nm respectively by Scherrer equation [19] from the (220) re ection in the Pt fcc lattice, which is not consistent with the results from TEM.This is because of the interference of a very small amount of large-sized Pt-NPs contained in the Pt/CNT-Jet sample.The peak intensities of Pt/CNT-Jet in XRD patterns are weaker than that of Pt/C with the same amout of Pt, suggesting that CNT carriers have much better dispersion compared with C carrier.
From the above analysis, the formation mechanism of the Pt/CNT by free arc can be clari ed as follows.Under the thermal excitation of free arc, water on the surface of cathode vaporizes rapidly, as shown in Fig. 6a→Fig.6b.Then a uniform thin layer of PtCl 4 will attach to the surface of CNTs that have lost their moisture protection (Fig. 6b).While the free arc continues to heat CNT@PtCl 4 to realize the decomposition of PtCl 4 (Fig. 6c).But this process alone cannot completely decompose all the PtCl 4 on CNT.
At the same time, affected by the free arc heat, the water in the PtCl 4 solution vaporizes quickly, which will generate a strong pressure eld (Fig. 6d).According to the previous research [10] , the pressure eld is similar to the scale of the discharge laments (variations of free arc).And it has two functions.One is to provide a strong endogenous driving force for the dispersion of CNTs, the other is to blow off the discharge lament.A new discharge lament is regenerated elsewhere on the surface of cathode by a high electric eld.The whole process restarts from the Fig. 6a.
Moisture is very important in this process.It can make sure CNTs disperse smoothly and make the process Fig.a→Fig.dcomplete within a short time (about 0.1ms) [10] .It is because of this short thermal process that Pt-NPs are formed in very small and independent states instead of fusion and agglomeration.
A small plasma jet at the anode nozzle can be generated by passing compressed air through the gap of electrodes.It is because of the heat of the plasma and the strong reaction activity, especially the relatively longer action time that the remaining Pt ions on CNT are further reduced.In brief, Pt-NPs prepared by the plasma jet are basically equivalent to those prepared by the chemical method.Tab.2.Results for the ECSA (m 2 /g) and retention per thousand cycles.
The cyclic voltammograms for the Pt/CNT prepared with plasma jet and for commercial Pt/C are displayed in Fig. 7.The cyclic voltammograms of the Pt/C is also presented for comparison.The CV curves show three potential regions: the hydrogen adsorption/desorption region (-0.2-0.1V),double layer plateau region (0.1-0.5V) and the formation and reduction of surface Pt oxides (0.5-1.0V).All voltammograms display a well-de ned hydrogen adsorption/desorption region from − 0.2 − 0.1 V vs. SCE.
Compared with Pt/C, Pt/CNT possess a wide double layer region, which is because CNTs have a higher ability to store electric charges.
The ECSA for Pt/CNT and Pt/C in the rst cycle are 44.15 and 48.38m 2 /g respectively on the basis of a monolayer hydrogen adsorption charge of 0.21mC/cm 2 on polycrystalline Pt.The ECSA for Pt/CNT in this work is roughly equivalent to commercial Pt/C and other reported Pt/CNT (40.21-58.4m 2 /g [20][21] ).It is signi cantly better than the 22.1m 2 /g of the similar liquid plasma method [22] .Although Pt/CNT has a more uniform dispersion and smaller mean size of Pt-NPs, the larger amount of large particle size of Pt-NPs in Pt/CNT than that in the Pt/C's makes its ECSA lower.
Moreover, the accelerated durability test (ADT) implemented by CVs investigated the stability of Pt/CNT and Pt/C electrocatalysts by judging the change of ECSA.
It can be seen from Fig. 7 and Table 2 that the ECSA of two Pt electrocatalysts decreases with the increasing number of cycles.Nonetheless, Pt/CNT electrocatalysts showed the higher retention (68.46%) compared with Pt/C (33.30%) electrocatalysts in the rst 1000 cycles.That indicates Pt-NPs supported on CNT have higher durability than those on Vulcan XC-72 carbon black.The improved stability of Pt/CNT was attributed to a large number of defects caused by the high temperature effect of free arc on the surface of CNT, which forms a strong interface function between Pt-NP and CNT.As well as CNTs carriers have better oxidation resistance and are less susceptible to etching during redox reactions that cause Pt-NP to detach from the carrier, so Pt/CNT has higher durability [23] .

4.
In summary, the direct action of the free arc can disperse CNTs well, but it cannot completely reduce the PtCl 4 on the surface of CNT.The dispersed CNTs carrying Pt ions leave the cathode and enter the plasma jet where all Pt ions can be reduced completely.And the mean size of the Pt-NPs prepared by the plasma jet method is 3.41nm, which is smaller than 4.77nm of the commercial catalyst.The ECSA for Pt/CNT is roughly equal to commercial Pt/C.ADT shows that Pt/CNT has better durability, which implies that CNT and Pt-NP have a stronger interaction under the action of a free arc.It is hoped that this method can be applied to the preparation of other elements and multi-element alloy catalysts in the future.

Declarations
This manuscript is submitted without any con ict of interest and all authors have approved the manuscript.I wish to state on behalf of my co-authors that the works described are original studies which have not been published and contemplated for publication elsewhere, wholly or partially.

Figures
Figures