chemical vapor deposition of highly dispersed pt nanoparticles on multi-walled carbon nanotubes for...

11
Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes Heeyeon Kim a, * , Sang Heup Moon b a Reaction and Separation Materials Research Center, Korea Institute of Energy Research, Jang-dong 71-2, Yuseong-gu, Daejeon 305-343, Republic of Korea b School of Chemical and Biological Engineering, Seoul National University, San 56-1, Shillim-dong, Kwanak-ku, Seoul 151-744, Republic of Korea ARTICLE INFO Article history: Received 6 May 2010 Accepted 7 December 2010 Available online 15 December 2010 ABSTRACT Fuel-cell electrode catalysts with improved electrochemical properties have been prepared by dispersing Pt nanoparticles onto carbon nanotubes (CNT) using a chemical vapor depo- sition (CVD) method. (Trimethyl)methylcyclopentadienyl platinum (MeCpPtMe 3 ) has been used as a Pt precursor in the CVD process and the CVD conditions have been optimized to disperse small Pt particles onto the CNT. Pt particles synthesized by CVD have a rela- tively uniform size of approximately 1 nm, which is substantially smaller than in the case of a commercial Pt/carbon black catalyst (64.5 nm) prepared by wet impregnation. The dis- persion of Pt, estimated by CO chemisorption, is also more than 14% greater than the com- mercial catalyst with these smaller particles. The electrochemically active surface area (ESA), measured by cyclic voltammetry (CV), and the long-time durability of the surface area of Pt/CNT prepared by CVD are higher than those of the commercial catalyst. Consequently, the single cell performance of the former catalyst is superior to that of the latter one. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction Since the discovery of carbon nanotubes (CNT) by Iijima in 1991 [1], their synthesis and applications have received increasing attention. Their unique properties, such as electri- cal conductivity, mechanical strength, and large surface area, give CNT the potential to be used in a number of applications, including conductive and high-strength composites [2], en- ergy-conversion devices [3], field-emission displays and radi- ation sources [4], batteries [5] and sensors [6,7]. So far, numerous studies [8–12] have investigated applying CNT to the preparation of catalysts for fuel-cell electrodes. For example, Matsumoto et al. [10] synthesized Pt particles rang- ing in size from 2 to 4 nm, and supported them with CNT in an attempt to reduce the amount of Pt in fuel cells’ electrodes. An electrode with 12 wt% Pt-deposited CNT produced 10% higher voltage than a 29 wt% Pt/carbon black electrode, and reduced the Pt requirement by 60% in polymer electrolyte fuel cells (PEFC). These results were suggested to be due to (i) well- dispersed Pt particles on the CNT surfaces, (ii) easy gas diffu- sion through networks and interiors of CNT, and (iii) the high electric conductivity of CNT. The surface of CNT, which consists of sp 2 -hybridized car- bons arranged in concentric graphene sheets, is highly hydro- phobic such that metal particles supported on them tend to agglomerate by van der Waals forces. Accordingly, oxidative treatment is necessary to create polar functional groups on the surface of CNT [13]. For example, Guha et al. [11] used 0008-6223/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2010.12.020 * Corresponding author: Fax: +82 42 860 3133. E-mail address: [email protected] (H. Kim). CARBON 49 (2011) 1491 1501 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/carbon

Upload: heeyeon-kim

Post on 21-Jun-2016

218 views

Category:

Documents


4 download

TRANSCRIPT

Page 1: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1

. sc iencedi rec t . com

ava i lab le a t www

journal homepage: www.elsevier .com/ locate /carbon

Chemical vapor deposition of highly dispersed Ptnanoparticles on multi-walled carbon nanotubesfor use as fuel-cell electrodes

Heeyeon Kim a,*, Sang Heup Moon b

a Reaction and Separation Materials Research Center, Korea Institute of Energy Research, Jang-dong 71-2, Yuseong-gu,

Daejeon 305-343, Republic of Koreab School of Chemical and Biological Engineering, Seoul National University, San 56-1, Shillim-dong, Kwanak-ku, Seoul 151-744,

Republic of Korea

A R T I C L E I N F O

Article history:

Received 6 May 2010

Accepted 7 December 2010

Available online 15 December 2010

0008-6223/$ - see front matter � 2010 Elsevidoi:10.1016/j.carbon.2010.12.020

* Corresponding author: Fax: +82 42 860 3133E-mail address: [email protected] (H. Ki

A B S T R A C T

Fuel-cell electrode catalysts with improved electrochemical properties have been prepared

by dispersing Pt nanoparticles onto carbon nanotubes (CNT) using a chemical vapor depo-

sition (CVD) method. (Trimethyl)methylcyclopentadienyl platinum (MeCpPtMe3) has been

used as a Pt precursor in the CVD process and the CVD conditions have been optimized

to disperse small Pt particles onto the CNT. Pt particles synthesized by CVD have a rela-

tively uniform size of approximately 1 nm, which is substantially smaller than in the case

of a commercial Pt/carbon black catalyst (64.5 nm) prepared by wet impregnation. The dis-

persion of Pt, estimated by CO chemisorption, is also more than 14% greater than the com-

mercial catalyst with these smaller particles. The electrochemically active surface area

(ESA), measured by cyclic voltammetry (CV), and the long-time durability of the surface

area of Pt/CNT prepared by CVD are higher than those of the commercial catalyst.

Consequently, the single cell performance of the former catalyst is superior to that of the

latter one.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Since the discovery of carbon nanotubes (CNT) by Iijima in

1991 [1], their synthesis and applications have received

increasing attention. Their unique properties, such as electri-

cal conductivity, mechanical strength, and large surface area,

give CNT the potential to be used in a number of applications,

including conductive and high-strength composites [2], en-

ergy-conversion devices [3], field-emission displays and radi-

ation sources [4], batteries [5] and sensors [6,7].

So far, numerous studies [8–12] have investigated applying

CNT to the preparation of catalysts for fuel-cell electrodes. For

example, Matsumoto et al. [10] synthesized Pt particles rang-

ing in size from 2 to 4 nm, and supported them with CNT in

er Ltd. All rights reserved

.m).

an attempt to reduce the amount of Pt in fuel cells’ electrodes.

An electrode with 12 wt% Pt-deposited CNT produced 10%

higher voltage than a 29 wt% Pt/carbon black electrode, and

reduced the Pt requirement by 60% in polymer electrolyte fuel

cells (PEFC). These results were suggested to be due to (i) well-

dispersed Pt particles on the CNT surfaces, (ii) easy gas diffu-

sion through networks and interiors of CNT, and (iii) the high

electric conductivity of CNT.

The surface of CNT, which consists of sp2-hybridized car-

bons arranged in concentric graphene sheets, is highly hydro-

phobic such that metal particles supported on them tend to

agglomerate by van der Waals forces. Accordingly, oxidative

treatment is necessary to create polar functional groups on

the surface of CNT [13]. For example, Guha et al. [11] used

.

Page 2: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

1492 C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1

acid-treated graphitic or amorphous carbon as a Pt support

for proton exchange membrane fuel cells (PEMFC). They

deposited Pt on carbon by a colloidal synthesis method to ob-

tain Pt particles ranging from 2 to 5 nm.

The corrosion of carbon support in an acidic solution is

one of the critical factors contributing to the degradation of

fuel-cell performance. As CNT have larger fractions of the

graphite component, they usually show higher resistance

against corrosion than in the case of carbon black (Vulcan

XC-72) [14]. It was also reported that Pt particles supported

on CNT showed stronger metal–support interactions than

those on Vulcan XC-72, accordingly the former exhibited bet-

ter performance as fuel cell catalyst [15].

Minimizing the cost of fuel-cell catalysts is highly desir-

able and the synthesis of nano-sized Pt particles is one of

ways to reduce the Pt loading without losing the catalytic

activity. Chemical vapor deposition (CVD), which involves

reactions between the surface sites of catalyst carriers and

vapors of metal compounds, is a useful method for preparing

nano-sized metal particles [12]. However, CVD rarely has been

used to prepare Pt nanoparticles, although it frequently is

used to obtain Pt thin films for electronic and optical devices

[16–18].

This is an advanced study of our previous work [19], which

focused on the different CVD conditions for the synthesis of

Pt nanoparticles of low metal loadings (�5 wt% Pt). The objec-

tives of present study were to synthesize highly dispersed and

nano-sized Pt particles onto CNT using a CVD method and to

investigate the potential of prepared catalysts as electrodes of

a fuel cell by comparing the electrochemical performance of

the prepared catalyst with that of commercial Pt/C. The Pt

contents of prepared catalysts were determined by induc-

tively coupled plasma-optical emission spectrometry (ICP-

OES), the dispersion of Pt particles was examined by trans-

mission electron microscopy (TEM), the number of catalytic

sites was estimated by CO chemisorption, and cyclic voltam-

metry (CV) and single cell test were performed for the estima-

tion of electrochemical activity.

2. Experiments

2.1. Materials

Table 1 lists the mechanical properties of the CNTused in this

study (MWNTs; CM-95, Iljin Nanotech). (Trimethyl)methylcy-

clopentadienyl platinum (MeCpPtMe3) (99%, Strem) was used

as a Pt precursor for CVD. For comparison, carbon black (Vul-

can XC-72) was also used as a support of Pt in CVD synthesis.

2.2. Pre-treatment of CNT

CNTwere annealed thermally at 470 �C for 50 min in air to re-

move carbonaceous particles. They were then immersed in

Table 1 – Properties of CNT used in this study (CM-95, Iljin Nan

Diameter (nm) Length (lm) Purity (wt%) Bulk de

CM-95 10–15 10–20 95

6 M hydrochloric acid (HCl) for 24 h to remove transition met-

als that might have become associated with the CNT during

synthesis; this process was repeated several times until the

acidic solution no longer changed color. Each sample was

washed several times with deionized water and dried at

110 �C for 12 h. The CNT obtained were refluxed in a mixture

of 14 M nitric acid (HNO3, 50 ml) and sulfuric acid (98% H2SO4,

50 ml) at 60 �C for 15 min–4 h. This treatment increased their

functionalization by creating defects on the surface, as dis-

cussed in Section 3.1. The final suspension was washed with

deionized water and filtered by a polytetrafluoroethylene

(PTFE) membrane (0.5 lm, Whatman), prior to drying at

110 �C for 12 h. Carbon black (Vulcan XC-72) was also treated

by the acid solution prior to Pt CVD following the same proce-

dure described above.

2.3. Catalyst synthesis

Fig. 1 shows the CVD apparatus used in this study. The Pt pre-

cursor was loaded into an evaporator and heated to 60 �C in

flowing N2 carrier gas. The optimal vaporization temperature

for the Pt precursor was determined by thermogravimetric

analysis (TGA) of MeCpPtMe3, as described in Section 3.2.

The vaporized precursor was transferred to a CVD reactor

containing 0.2 g of pre-treated CNT. The reactor temperature

was raised from 80 to 300 �C while its pressure was main-

tained at 6–10 Torr. For the CVD of Pt, various gases were

added to the N2 carrier gas (20 sccm), including N2 (20 sccm),

O2 (20% O2/Ar, 20 sccm), and H2 (5% H2/Ar, 20 sccm). Pt/CNT

catalysts synthesized by above process were indicated as

‘Pt/CNT (CVDX/Y)’, X and Y denoting the CVD temperature

in �C and the gas used for CVD, respectively. For comparison,

Pt/carbon black was also prepared by the same process de-

scribed above and indicated as ‘Pt/C (CVDX/Y)’, X and Y

meaning the same parameter as in the case of Pt/CNT. Com-

mercial Pt/carbon black (Hispec�4000, Pt 40 wt% , Johnson

Matthey) was used as a model catalyst and denoted as ‘Pt/C

(com)’.

2.4. Surface characterization and electrochemicalmeasurement

TGA of the Pt precursor was performed using a Q50 analyzer

(TA instruments). The Pt content of the catalyst was mea-

sured by inductively coupled plasma-optical emission spec-

trometry (ICP-OES, Perkin-Elmer). The amounts of CO

chemisorbed onto Pt were measured using an Autochem

2910 (Micromeritics). The surface morphology of each catalyst

was also examined by high resolution transmission electron

microscopy (HR-TEM, JEOL JEM).

The electrochemically active surface area (ESA) of each

catalyst was estimated by CV analysis using a potentiostat

(Biologic sp-50). The working electrode preparation was

otech).

nsity (g/cc) Oxidation temp. (�C) BET surface area (m2/g)

0.1 600 200

Page 3: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

Fig. 1 – CVD apparatus.

C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1 1493

similar to the process described by Schmidt et al. [20]; 30 mg

of the supported catalyst was ultrasonically blended with

60 ll deionized water, 979.2 ll of isopropyl alcohol and

244.8 ll of 5 wt% Nafion solution (Dupont Fluoroproducts)

for 20 min. Then, 2.5 ll of the catalyst slurry was pipetted

onto a 3 mm diameter glassy carbon disk electrode (glassy

carbon area: 0.0706 cm2) and the electrode was dried in air.

Pt loading of the resultant glassy carbon disk was

0.33 mg Pt cm�2, when the catalyst contained 40 wt% Pt. The

glassy carbon disk electrode was then immersed in N2 deaer-

ated electrolyte (0.5 M H2SO4 aqueous solution) and used as a

working electrode at 25 �C. A reversible hydrogen electrode

and Pt wire were used as reference and counter electrodes,

respectively. The applied potential was repeated in cycles in

the range of 0–1.4 V and compared to a normal hydrogen elec-

trode (NHE) at a scan rate of 20 mV s�1. All electrochemical

experiments were carried out at room temperature.

The single cell performance of each catalyst was tested

using either Pt/CNT prepared by CVD or commercial Pt/C

(40 wt% Pt) as a cathode and commercial Pt/C (40 wt% Pt) as

an anode at 70 �C in H2/O2 system. The membrane-electrode

assembly (MEA) was prepared using Nafion as a polymer elec-

trolyte. The catalyst loading in the cathode and anode was

0.25 mgPt cm�2 and 0.3 mgPt cm�2, respectively. The single cell

with active area of 5 cm2 was fed with 200 ml min�1 of pure O2

at the cathode and with 200 ml min�1 of H2 at the anode.

The samples were also analyzed by X-ray photoelectron

spectroscopy (XPS) using a MultiLab 2000 (Thermo Electron

Co.) with an Mg Ka X-ray source.

3. Results and discussion

3.1. XPS results of pre-treated CNT

The extents of surface functionalization of CNT samples pre-

treated under different conditions were measured by XPS.

Fig. 2 shows the XPS of C and O, and Table 2 presents the

atomic percentages of various elements as determined from

the XPS spectral areas. Fig. 2 shows little difference in the

C1s signals between the acid treated and untreated samples.

However, substantial differences were observed in the O1s

signals according to the pre-treatment conditions, i.e., the

peak area increased in proportion to the treatment time, as

summarized Table 2. These results mean that the acid treat-

ment was effective for generating surface defects, which were

required for the deposition and dispersion of metal nanopar-

ticles. In addition, the O1s peak was slightly shifted to higher

binding energies with an increase in the acid-treatment peri-

od. As the literature [21] suggested, the O1s peak of acid-trea-

ted CNT was actually composed of two peaks (532.0 and

533.5 eV) caused by carbon oxidation. Fig. 2 shows the decon-

volution results of O1s peak in the cases of (a) and (f), which

indicate that the area ratio of the peak at 532.0 eV, corre-

sponding to C–O, over one at 533.5 eV, corresponding to

C@O, was decreased by the acid treatment. Accordingly, the

surface treatment produced larger amounts of C@O than

the amounts of C–O. In all cases, the peaks for S2p and N1s

were negligible, which suggested that the acid solution was

nearly completely removed from the CNT samples after the

washing process.

3.2. Pt-CVD on CNT

The thermal behavior of MeCpPtMe3 was studied by TGA

(Fig. 3) to determine the optimal temperature for vaporizing

the Pt precursor for the CVD process. Vaporization was initi-

ated at approximately 50 �C and was completed at 150 �C in

both N2 and O2 atmospheres. Accordingly, a temperature of

60 �C was used for all subsequent experiments. The TGA of

MeCpPtMe3 could not be performed in H2 because the precur-

sor rapidly decomposed upon exposure to H2, even at room

temperature.

Page 4: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

294 292 290 288 286 284 282 280 278

Inte

nsity

(A

.U.)

Binding energy (eV)

(f)

(e)

(d)

(c)

(b)

(a)

C1s

538 536 534 532 530 528 526

O1s

(f)

(e)

(d)

(c)

(b)

Binding energy (eV)In

tens

ity (

A.U

.)

(a)

Fig. 2 – The XPS spectra of C1s and O1s with acid-pre-treatment time; (a) pristine CNT, (b) 15 min, (c) 30 min, (d) 1 h, (e) 2 h and

(f) 4 h. The peak deconvolution of O1s spectrum is shown in the cases of (a) and (f).

Table 2 – The atomic percentage of the various elements at the CNT sample surfaces.a

Samples Time (min)b C% O% S% N%

Pristine 0 98.32 1.68 – –Sample A 15 97.54 2.46 – –Sample B 30 97.28 2.72 – –Sample C 60 97.25 2.75 – –Sample D 120 97.02 2.98 – –Sample E 240 96.23 3.66 0.11 –a The atomic percentage was measured from XPS spectra.b Each samples were treated by mixed acid solution of 14 M nitric acid (HNO3, 50 ml) and sulfuric acid (98% H2SO4, 50 ml) at 60 �C.

0 50 100 150 200 250 300

0

20

40

60

80

100

(b)

Wei

ght l

oss

(%)

Temperature ( oC)

(a)

Fig. 3 – TGA of (trimethyl)methylcyclopentadienyl platinum

(MeCpPtMe3) in (a) N2 and (b) O2 atmospheres; temperature

was increased at a rate of 5 �C/min from 30 �C to 500 �C.

1494 C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1

Previous studies have utilized a number of conditions for

the CVD preparation of Pt films or particles, using MeCpPtMe3

as a Pt precursor. For example, Aaltonen et al. [16] obtained a

Pt thin film with excellent uniformity, purity and low resis-

tance by atomic layer deposition (ALD) in O2 at 300 �C onto

borosilicate glass and onto a silicon substrate. Hiratani et al.

[17] used Ar as a carrier gas to evaporate MeCpPtMe3 and O2

as a reactive gas in the CVD process, at temperatures between

200 and 400 �C. The thermal decomposition of the Pt precur-

sor dominated at a low O2 ratio (O2/Ar = 1/20), while the oxi-

dative decomposition of the precursor controlled the film

growth rate at high O2 ratios (O2/Ar = 2/1 and 7/1). Valet

et al. [22] synthesized a Pt thin film by the oxidative decompo-

sition of MeCpPtMe3 diluted in cyclohexane. Lashdaf et al. [23]

prepared a Pt catalyst supported on alumina or silica by the

gas-phase deposition of Pt at reduced pressures, using MeC-

pPtMe3 as a Pt precursor and nitrogen as a carrier gas. The

Pt precursor was vaporized at 60 �C and allowed to react with

a fixed-bed support that had been stabilized at 100 �C.

In the present study, Pt particles were deposited onto CNT

using different carrier gases and temperatures in order to

determine the optimal deposition condition. N2 was used as

a carrier gas to transfer the Pt precursor from the evaporator

Page 5: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

50 100 150 200 250 300

0

10

20

30

40

(c)

(b)

Pt l

oadi

ng (

wt%

)

Temperature (oC)

(a)

Fig. 4 – Changes in the Pt loading of Pt/CNT catalysts

prepared under different conditions: (a) in 100% N2, (b) in

10% O2 and (c) in 2.5% H2.

C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1 1495

to the CVD reactor, and either N2, O2 or H2 was added to the

carrier gas flowing into the CVD reactor.

Fig. 4 indicates that the rates of Pt-CVD depended on both

the reaction temperature and the atmospheric gas. In N2 (case

(a)), the Pt loading was insignificant and even slightly de-

creased at temperatures below 140 �C, but increased sharply

between 140 and 200 �C, with a slow rate of increase as the

temperature approached 300 �C. The initial decrease at tem-

peratures below 140 �C is possibly due to the increased

desorption of the precursor from the substrate. The rates of

Pt-CVD were accelerated above 140 �C because the precursor’s

decomposition on the substrate exceeded the precursor’s

desorption from the surface. The slow rate of increase in

CVD above 200 �C was due to the partial decomposition of

the Pt precursor in the gas phase. Nevertheless, Pt loading

as high as 41.7 wt% was still achieved at 300 �C.

When O2 was added to the CVD reactor at an O2/N2 ratio of

1/2, Pt deposition proceeded at lower temperatures than had

been seen in the N2 atmosphere. Pt-CVD apparently was facil-

itated by the Pt precursor’s oxidative decomposition, as Hira-

tani et al. [17] reported previously. However, at temperatures

over 150 �C, the Pt loading was lower than that obtained in

N2 because O2 promoted both the decomposition and desorp-

tion of the precursor.

The Pt precursor, MeCpPtMe3, rapidly decomposed upon

contact with H2, showing a sharp exothermic peak in the

TGA (data not shown). During CVD in the presence of H2,

the deposition of Pt on CNT was minimal. On the other hand,

Pt fragments produced in the gas phase immediately attached

to the SiO2 walls of the CVD reactor to form a Pt thin film.

Including H2 in the carrier gas was therefore detrimental to

the CVD of Pt nanoparticles onto CNT.

3.3. Surface characterization

3.3.1. TemFig. 5 shows TEM images of the Pt particles deposited onto

CNT in N2 at different temperatures. In the case of Pt/CNT

(CVD100/N2) (Fig. 5(a)), fine Pt particles with diameters smaller

than 1 nm were dispersed onto the support, at a Pt loading of

12.2 wt% in this case (Fig. 4). In the case of Pt/CNT (CVD200/

N2) (Fig. 5(b)), the Pt loading increased to 32 wt% (Fig. 4) with

little change in the size of the Pt particles. The increase in

Pt loading was therefore due primarily to an increase in the

number of Pt particles deposited. A similar trend was ob-

served when the temperature was raised to 300 �C, that is,

the increased Pt loading (41.7%) of Pt/CNT (CVD300/N2)

(Fig. 5(c)) again could be attributed to an increase in the num-

ber of particles, not the size of them. Overall, Pt particles of a

uniform size, about 1 nm, were dispersed onto CNT with a Pt

loading of ca. 5–40 wt% by CVD in N2 by varying the deposition

temperature. The dispersed Pt particles apparently were ad-

sorbed selectively onto the defects in the CNT surface gener-

ated by acid pre-treatment [10].

Fig. 6 shows TEM images of Pt/CNT (CVD120/O2) on differ-

ent scales of magnification. As seen in Fig. 4, Pt particles were

produced at lower temperatures in an O2 than in an N2 atmo-

sphere. The Pt loading of Pt/CNT (CVD120/O2) was as high as

ca. 35 wt% (Fig. 4). However, compared to the particles pro-

duced in N2, the Pt particles obtained in O2 (Fig. 6) were not

uniform in shape and were much larger, ranging from 5 to

30 nm. Therefore, the presence of O2 in the carrier gas facili-

tated the formation of Pt clusters or films, as described in pre-

vious studies [17,18,22], rather than the formation of

nanoparticles.

TEM images of Pt/C (CVD300/N2) (Pt 40.3 wt%) were also

obtained and shown in Fig. 7. Similar to the case of Pt/CNT

in Fig. 5, Pt particles are highly dispersed on carbon black

and the size of them is as small as 1 nm.

Fig. 8 shows TEM images of a Pt/C (com) (Pt 40.0 wt%), in

which the Pt particles are non-uniform in shape and in the

size range of 2–5 nm, which is larger than in both catalysts

prepared by CVD (Fig. 5 and Fig. 7). Consequently, the best

Pt nanoparticles were produced by CVD in N2.

Fig. 9 shows XRD patterns for Pt/CNT (CVD300/N2) (Pt

41.7 wt%), Pt/C (CVD300/N2) (Pt 40.3 wt%), and Pt/C (com) (Pt

40.0 wt%). Based on these results, the Pt particle size of indi-

vidual catalyst was measured according to the Scherrer equa-

tion. As shown in Table 3, the average particle size calculated

shows values similar to those measured from CO chemisorp-

tion data.

3.3.2. CO chemisorptionTable 3 lists the surface areas, dispersion, and average

diameters of Pt particles obtained in four sample catalysts,

as estimated from the results of CO chemisorption. Although

the Pt loading was different for each catalyst, between 12.2

and 41.7 wt%, the Pt surface area per Pt mass was consistently

larger for those prepared by CVD than the commercial cata-

lyst. For example, the surface area based on the weight of Pt

of Pt/CNT (CVD100/N2) (Pt 12.2 wt%) was 1.4-fold larger than

that of the Pt/C (com) (Pt 40 wt%). Likewise, Pt/CNT

(CVD300/N2) (Pt 41.7 wt%) had a Pt surface area 1.2-fold larger

than that of the commercial catalyst. The CVD process al-

lowed much higher dispersion of the Pt particles, also a uni-

form size, compared with the case of using impregnation

for the commercial catalyst. Pt particles of Pt/C (CVD300/N2)

show similar values in Pt surface area, the dispersion and

the diameter of Pt particles to those of Pt/CNT. As shown in

Table 3, the high Pt dispersion of 75–83% obtained in the

CVD-prepared samples was preserved up to a Pt loading of

Page 6: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

Fig. 5 – TEM images of (a) Pt/CNT (CVD100/N2), (b) Pt/CNT (CVD200/N2) and (c) Pt/CNT (CVD300/N2).

Fig. 6 – TEM images of Pt/CNT (CVD120/O2) (Pt 35 wt%), shown on different scales of magnification.

1496 C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1

41.7 wt%, which was larger than 61% obtained in the case of

Pt/C (com). The average diameter of Pt particles estimated

from the CO chemisorption data (Table 3) corroborated what

the TEM images shown in Figs. 5, 7 and 8. ESA values are also

given in Table 3 for the discussion of Pt dispersion and elec-

trocatalytic performance, as given in the following section.

Page 7: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

Fig. 7 – TEM images of Pt/C (CVD300/N2) (Pt 40.3 wt%), obtained at different sites of the catalyst.

Fig. 8 – TEM images of Pt/C (com) (Pt 40.0 wt%), shown on different scales of magnification.

20 30 40 50 60 70 80

(c)

(b)

(220)(200)

Inte

nsity

(A

.U.)

2θ (degree)

(111)

(a)

*

Fig. 9 – Powder XRD patterns of (a) Pt/CNT (CVD300/N2) (Pt

41.7 wt%), (b) Pt/C (CVD300/N2) (Pt 40.3 wt%), and (c) Pt/C

(com) (Pt 40.0 wt%). Miller index of each diffraction signal is

denoted on the top of the signal and the broad carbon

background signal is marked by *.

C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1 1497

3.3.3. Cyclic voltammetryFor the evaluation of ESA and the investigation of electro-

chemical stability of various Pt/C catalysts, the catalysts were

subjected to 1000 cycles of CV and the results are shown in

Fig. 10.

All the catalysts in Fig. 10 show degradation in ESA with an

increase in the cyclic number, but the trends are not the same

among them. ESA was measured from the peak area in the

potential range of 0.01–0.35 V, corresponding to the adsorp-

tion/desorption of hydrogen, and the results are summarized

in Fig. 11. The ESA data shown in Fig. 11 designate the relative

values based on the ESA of Pt/C (com) at the initial stage of CV.

At the initial stage, the ESA of Pt/C (CVD300/N2) was 1.2-fold

larger than that of Pt/C (com). The reason is possibly due to

the increased Pt surface area obtained by CVD synthesis as

observed in our CO chemisorption results (Table 3). Moreover,

the ESA of Pt/CNT (CVD300/N2) was 1.5-fold larger than that of

Pt/C (com). The above results suggest that there is another

factor, in addition to the increased Pt surface area, that

Page 8: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

con

tribu

tes

toth

ein

crease

inE

SA

of

Pt/C

NT

cata

lyst.

Inth

is

resp

ect,th

eim

pro

ved

ele

ctrical

con

du

ctivity

of

CN

Tco

m-

pa

redto

tha

to

fca

rbo

nb

lack

can

be

con

sidere

da

sa

no

ther

rea

son

for

the

en

ha

nced

ele

ctroch

em

ical

perfo

rma

nce

of

Pt/

CN

Tca

taly

st.W

an

get

al.

[24]

repo

rted

tha

tth

eim

pro

ved

per-

form

an

ceo

fP

t/CN

Tele

ctrod

es

com

pa

red

toP

t/carb

on

bla

ck

wa

sd

ue

tola

rger

am

ou

nts

of

the

triple

-ph

ase

bo

un

da

ries

form

ed

on

the

Pt/C

NT

ele

ctrod

es,

wh

icha

llow

ed

the

hig

h

utiliza

tion

of

Pt

inth

eca

taly

st.T

he

incre

ase

of

ES

Afo

rth

e

Table 3 – Characterization of sample catalysts based on CO chemisorption.

Catalyst Pt loadinga

(wt%)Amount of

adsorbed CO(ml/g cat.)

Pt surface areab

(m2/g cat.)[relative values]

Pt surface areab

(m2/g Pt)[relative values]

Ptdispersionb (%)

Ave. particlediameterIb (nm)

Ave.particle

diameter IIc (nm)

ESAd

(relative values)

Pt/CNT (CVD100/N2)e 12.2 13.0 25.0 [0.4] 205.0 [1.4] 83 [1.4] 1.3 – –Pt/CNT (CVD300/N2)f 41.7 40.0 77.0 [1.3] 184.7 [1.2] 75 [1.2] 1.4 0.9 1.5Pt/C (CVD300/N2)g 40.3 38.7 74.5 [1.2] 178.7 [1.2] 76 [1.2] 1.4 1.0 1.2Pt/C (com)h 40.0 – 60.0 [1.0] 150.0 [1.0] 61 [1.0] 64.5 3.0 1.0

a By ICP-OES analysis (data for Pt/C (com) were given from Johnson Matthey).b Based on CO chemisorption data using 5 mg of catalyst at 30 �C (data for Pt/C (com) were given from Johnson Matthey).c Based on XRD results using the Scherrer equation (d ðnmÞ ¼ 0:9k

B cosðhÞ, d is the diameter of Pt particle, k is the X-ray wavelength (0.154 nm), B is the full width at half height for the diffraction peaks in

radians, and h is half of the diffraction angle).d Electrochemically active surface area at the beginning of CV test, determined from Fig. 10.e Pt/CNT (CVD100/N2) (CVD for 2 h).f Pt/CNT (CVD300/N2) (CVD for 2 h).g Pt/C (CVD300/N2) (CVD for 2 h).h Pt/C (com), data provided by Johnson Matthey.

0.00.2

0.40.6

0.81.0

1.21.4

-20

-15

-10 -5 0 5 10 15

Potential (V

) (vs. NH

E)

Current (mA/g Pt)

V IVIII

III

(a)

0.00.2

0.40.6

0.81.0

1.21.4

-20

-15

-10 -5 0 5 10 15

Current (mA/g Pt)

Potential (V

) (vs. NH

E)

V IV

IIIII

I

(b)

0.00.2

0.40.6

0.81.0

1.21.4

-20

-15

-10 -5 0 5 10 15

V IVIII

II

Potential (V

) (vs. NH

E)

Current (mA/g Pt)

I

(c)

Fig

.10

–C

yclic

vo

ltam

mo

gra

ms

of

(a)P

t/CN

T(C

VD

30

0/N

2 )(Pt

41

.7w

t%),

(b)

Pt/C

(CV

D3

00

/N2 )

(Pt

40

.3w

t%)

an

d(c)

Pt/C

(com

)(P

t4

0.0

wt%

),re

cord

ed

ina

0.5

MH

2 SO

4so

lutio

na

t

25

�Cin

N2 .

Po

ten

tial

scan

rate

wa

s2

0m

Vs�

1.E

ach

spectru

mw

as

ob

tain

ed

(I)a

tth

ein

itial

stag

eo

fC

V,

(II)a

fter

25

0cy

cles,

(III)a

fter

50

0cy

cles,

(IV)

afte

r7

50

cycle

s,a

nd

(V)

afte

r1

00

0cy

cles.

14

98

CA

RB

ON

49

(2

01

1)

14

91

–1

50

1

Page 9: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

0 250 500 750 10000.4

0.6

0.8

1.0

1.2

1.4

1.6

(c)

(b)

ES

A r

elat

ive

valu

es

Cycle (times)

(a)

Fig. 11 – Electrochemically active surface area (ESA)

evaluated from the cyclic voltammograms shown in Fig. 10:

(a) Pt/CNT (CVD300/N2), (b) Pt/C (CVD300/N2), and (c) Pt/C

(com). ESA data designate relative values based on that of

Pt/C (com) at the initial stage.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.40.0

0.2

0.4

0.6

0.8

1.0

0.0

0.1

0.2

0.3

0.4

(b)

(a)

(b)

Pow

er d

ensi

ty

W c

m-2

Cel

l vol

tage

(V

)

Current density A cm-2

(a)

Fig. 12 – Single cell performance of (a) Pt/CNT (CVD300/N2)

(Pt 41.7 wt%) and (b) Pt/C (com) (Pt 40.0 wt%).

80 78 76 74 72 70

(a)

Pt4f7/2

(d)

(c)

(b)Inte

nsity

(A

.U.)

Binding energy (eV)

Pt4f5/2

Fig. 13 – The Pt 4f core level XPS spectra of (a) Pt/C (com)

(Pt 40.0 wt%), (b) Pt/C (CVD300/N2) (Pt 40.3 wt%), (c) Pt/CNT

(CVD300/N2) (Pt 41.7 wt%), and (d) Pt/CNT (CVD120/O2)

(Pt 35 wt%).

C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1 1499

CNT-supported electrodes, observed in this study, can be ex-

plained based on the same scenario as the above.

Meanwhile, all the catalysts showed a decrease in ESA

with an increase in the cyclic number, which could be attrib-

uted to several factors such as crystal growth, agglomeration

of catalyst particles, and carbon corrosion [25]. As Xie et al.

mentioned previously [26], smaller crystallites have higher

surface tension and driving force for assembling themselves

into larger particles to reduce the overall surface energy of

the total catalyst mass. Additionally, the weak bonding of cat-

alyst particles with the carbon support on a fresh catalytic

layer can facilitate the agglomeration of the nanoparticles in

the initial stage of operation. Fig. 11 shows that the ESA val-

ues of Pt/C catalysts prepared by CVD, Fig. 11(a) and (b), are

decreased to a little greater extent during the initial 250 cycles

of operation compared to the case of commercial catalyst

(Fig. 11(c)). The reason is because the Pt particles of the former

two catalysts are smaller than those of the latter one, such

that the Pt particles agglomerate at higher rates in the former

catalysts. Nevertheless, the ESA values of the former catalysts

are still larger than that of the latter even after 1000 cycles

of operation. In fact, the ESA of Pt/CNT (CVD300/N2) is still

1.5-fold larger and that of Pt/C (CVD300/N2) is 1.2-fold larger

than that of Pt/C (com) after 1000 cycles. The reason can be

because the increased number of catalytic sites is maintained

even after long-time operation due to strong metal–support

interactions caused by the reduced size of Pt particles.

Single cell performance of Pt/CNT (CVD300/N2) (Pt

41.7 wt%) was measured by using H2/O2 system at 70 �C and

the result was compared with that of Pt/C (com) (Pt

40.0 wt%). Fig. 12 indicates that the single cell performance

of the former catalyst is superior to that of the latter, which

is mainly due to the highly dispersed Pt nanoparticles on

CNT and the unique 3D structure of the CNT-based electrode

which improves the mass transport [27].

We investigated the metal–support interactions of various

types of catalysts using XPS, and the results are shown in the

following section.

3.3.4. XpsFig. 13 shows the XPS spectra of Pt 4f for several catalysts

of this study. The commercial catalyst (Fig. 13(a)) displayed

an intense peak at 71.6 eV, which is assigned as a 4f7/2 signal

of zero-valent Pt. The other peak at 74.7 eV is assigned as a

4f5/2 signal. In both cases of CVD-prepared catalysts

(Fig. 13(b) and (c)), which show similar trends in XPS results,

the spectra show a shift to higher binding energies (72.1 and

75.2 eV) than for the commercial Pt/C catalyst (Fig. 13 (a)).

These results are attributed to the metal–support interactions

Page 10: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

1500 C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1

between small Pt particles of the CVD-prepared catalyst and

the support, either CNT or carbon black, compared with the

case of the commercial catalyst [28–30]. On the contrary, the

spectra of Pt/CNT (CVD120/O2) (Fig. 13(d)) showed a small shift

to lower binding energies (71.3 and 74.5 eV), apparently due to

weak metal–support interactions in the catalyst containing

large Pt particles. Consequently, the increased Pt–support

interactions obtained with small Pt particles prepared by

CVD in N2 contributed beneficially to the ESA and the long-

term durability of the fuel-cell catalysts, as confirmed by

our CV results in Section 3.3.3.

4. Conclusion

Pt/CNT catalysts were synthesized by depositing Pt onto CNT

under various CVD conditions and were then characterized

using XPS, TEM, and CO chemisorption. The electrochemi-

cally active surface area (ESA) of the prepared catalysts, esti-

mated by cyclic voltammetry (CV), was compared with those

of Pt/carbon black prepared by CVD and a commercial Pt/C

catalyst.

Pt was successfully dispersed onto CNT as particles of a

uniform size, ca. 1 nm, by CVD in an N2 atmosphere. The high

dispersion of Pt particles, in the range of 75–83%, was pre-

served even when the Pt loading increased from 12.2 to

41.7 wt% by Pt-CVD at elevated temperatures of 100–300 �C.

The resulting Pt/CNT catalysts prepared by CVD had signifi-

cantly larger number of catalytic sites, i.e., Pt surface area

per Pt mass, than the commercial Pt/C catalyst. The former

catalysts show larger ESA than the latter one even after

1000 cycles of CV test. The CVD preparation technique de-

scribed herein can improve the stability of Pt/C catalysts be-

cause of the enhanced Pt–support interactions obtained

with small Pt particles.

The ESA of Pt/CNT prepared by CVD was larger than that of

Pt/C prepared under the same CVD condition, even when the

average diameter of Pt particles was the same between two

catalysts, which indicated that an additional factor besides

the Pt particle size contributed to the ESA enhancement in

the CNT-supported catalyst. It has been proposed in this

study that the higher electrical conductivity of CNT compared

to that of carbon black is responsible for the enhanced ESA of

Pt/CNT prepared by CVD.

Consequently, the single cell performance of Pt/CNT

(CVD300/N2) (Pt 41.7 wt%) is superior to that of Pt/C (com)

(Pt 40.0 wt%), which is due to the highly dispersed Pt nanopar-

ticles on CNT and the unique 3D structure of the CNT-based

electrode.

R E F E R E N C E S

[1] Iijima S. Helical microtubules of graphitic carbon. Nature(London) 1991;354:56–8.

[2] Qian D, Dickey EC, Andrews R, Rantell T. Loadtransfer and deformation mechanisms in carbonnanotube–polystyrene composites. Appl Phys Lett2000;76:2868–70.

[3] Niu C, Sickel EK, Hoch R, Moy D, Tennent H. High powerelectrochemical capacitors based on carbon nanotubeelectrodes. Appl Phys Lett 1997;70:1480–2.

[4] de Heer WA, Chatelain A, Ugarte D. A carbon nanotube field-emission electron source. Science 1995;270:1179–80.

[5] Che G, Lakshmi BB, Martin CR, Fisher ER. Metal-nanocluster-filled carbon nanotubes: catalytic properties and possibleapplications in electrochemical energy storage andproduction. Langmuir 1999;15(3):750–8.

[6] Britto PJ, Santhanam KSV, Ajayan PM. Carbon nanotubeelectrode for oxidation of dopamine. BioelectrochemBioenerg 1996;41:121–5.

[7] Davis JJ, Coles RJ, Hill HAO. Protein electrochemistry at carbonnanotube electrodes. J Electroanal Chem 1997;440:279–82.

[8] Cui SK, Guo DJ. Highly dispersed Pt nanoparticlesimmobilized on 1,4-benzenediamine-modified multi-walledcarbon nanotube for methanol oxidation. J Colloid Interf Sci2009;333:300–3.

[9] Lebert M, Kaempgen M, Soehn M, Wirth T, Roth S, Nicoloso N.Fuel cell electrodes using carbon nanostructures. Catal Today2009;143:64–8.

[10] Matsumoto T, Komatsu T, Arai K, Yamazaki T, Kijima M,Shimizu H, et al. Reduction of Pt usage in fuel cellelectrocatalysts with carbon nanotube electrodes. ChemCommun 2004:840–1.

[11] Guha A, Lu W, Zawodzinski TA, Schiraldi DA. Surface-modified carbons as platinum catalyst support for PEM fuelcells. Carbon 2007;45(7):1506–17.

[12] Inumaru K, Okuhara T, Misono M. Elementary surfacereactions in the preparation of vanadium oxide overlayers onsilica by chemical vapor deposition. J Phys Chem1991;95(12):4826–32.

[13] Obradovic MD, Vukovic GD, Stevanovic SI, Panic VV,Uskokovic PS, Kowal A, et al. A comparative study of theelectrochemical properties of carbon nanotubes and carbonblack. J Electroanal Chem 2009;634:22–30.

[14] Wang X, Li W, Chen Z, Waje M, Yan Y. Durabilityinvestigation of carbon nanotube as catalyst support forproton exchange membrane fuel cell. J. Power Sources2006;158:154–9.

[15] Luo JZ, Gao LZ, Leung YL, Au CT. The decomposition of NO onCNT and 1 wt% Rh/CNT. Catal Lett 2000;66:91–7.

[16] Aaltonen T, Ritala M, Sajawaara T, Keinonen J, Leskela M.Atomic layer deposition of platinum thin films. Chem Mater2003;15:1924–8.

[17] Hiratani M, Nabatame T, Matsui Y, Kimura S. Crystallographicand electrical properties of platinum film grown by chemicalvapor deposition using(methylcyclopentadienyl)trimethylplatinum. Thin Solid films2002;410:200–4.

[18] Thurier C, Doppelt P. Platinum OMCVD processes andprecursor chemistry. Coordin Chem Rev 2008;252:155–69.

[19] Kim H, Lee SJ, Jeong NJ. Chemical vapor deposition of Pt nanoparticles on carbon nanotubes. Solid State Phenom 2007;124–126:1769–72.

[20] Schmidt TJ, Gasteiger HA, Stab GD, Urban PM, Kolb DM, BehmRJ. Characterization of high-surface-area electrocatalystsusing a rotating disk electrode configuration. J ElectrochemSoc 1998;145:2354–8.

[21] Zhang G, Sun S, Yang D, Dodelet J-P, Sacher E. The surfaceanalytical characterization of carbon fibers functionalized byH2SO4/HNO3 treatment. Carbon 2008;46(2):196–205.

[22] Valet O, Doppelt P, Baumann PK, Schumacher M, Balnois E,Bonnet F, et al. Study of platinum thin films deposited byMOCVD as electrodes for oxide applications. MicroelectronEng 2002;64:457–63.

[23] Lashdaf M, Lahtinen J, Lindblad M, Venalainen T, Krause AOI.Platinum catalysts on alumina and silica prepared by gas-

Page 11: Chemical vapor deposition of highly dispersed Pt nanoparticles on multi-walled carbon nanotubes for use as fuel-cell electrodes

C A R B O N 4 9 ( 2 0 1 1 ) 1 4 9 1 – 1 5 0 1 1501

and liquid-phase deposition in cinnamaldehydehydrogenation. Appl Catal 2004;276:129–37.

[24] Wang JJ, Yin GP, Zhang J, Wang ZB, Gao YZ. High utilizationplatinum deposition on single-walled carbon nanotubes ascatalysts for direct methanol fuel cell. Electrochim Acta2007;52:7042–50.

[25] Prasanna M, Cho EA, Lim TH, Oh IH. Effectsof MEA fabrication method on durability of polymerelectrolyte membrane fuel cells. Electrochim Acta2008;53:5434–41.

[26] Xie J, Wood DL, More KL, Atanassov P, Borupa RL.Microstructural changes of membrane electrode assembliesduring PEFC durability testing at high humidity conditions. JElectrochem Soc 2005;152:A1011–20.

[27] Saha MS, Li R, Sun X. High loading and mono dispersed Ptnanoparticles on multiwalled carbon nanotubes for highperformance proton exchange membrane fuel cells. J PowerSources 2008;177:314–22.

[28] Sen F, Gokagac G. Different sized platinum nanoparticlessupported on carbon: an XPS study on these methanoloxidation catalysts. J Phys Chem C 2007;111:5715–20.

[29] Lim DH, Choi DH, Lee WD, Lee HI. A new synthesis of a highlydispersed and CO tolerant PtSn/C electrocatalyst for low-temperature fuel cell; its electrocatalytic activity and long-term durability. Appl Catal B: Environ 2009;89:484–93.

[30] Roth C, Goetz M, Fuess H. Synthesis and characterization ofcarbon-supported Pt–Ru–WOx catalysts by spectroscopic anddiffraction methods. J Appl Electrochem 2001;31:793–8.