Supporting Information for
Facet-Dependent Catalytic Activity of Platinum Nanocrystals for Triiodide Reduction in Dye-Sensitized Solar Cells Bo Zhang1, 2†, Dong Wang3†, Yu Hou1, Shuang Yang1, Xiao Hua Yang1, Ju Hua Zhong2, Jian Liu4, Hai Feng Wang3*, P. Hu3, 5, Hui Jun Zhao6, Hua Gui Yang1, 6* 1
Key Laboratory for Ultrafine Materials of Ministry of Education, School of Materials
Science and Engineering, East China University of Science & Technology, 130 Meilong Road, Shanghai 200237, China. 2
Department of physics, East China University of Science & Technology, 130 Meilong
Road, Shanghai 200237, China. 3
State Key Laboratory of Chemical Engineering, Centre for Computational Chemistry
and Research Institute of Industrial Catalysis, East China University of Science & Technology, 130 Meilong Road, Shanghai 200237, China. 4
ARC Centre of Excellence for Functional Nanomaterials, Australian Institute for
Bioengineering and Nanotechnology, The University of Queensland, QLD, 4072, Australia 5
School of Chemistry and Chemical Engineering, The Queen’s University of Belfast,
Belfast, BT9 5AG, UK. 6
Centre for Clean Environment and Energy, Gold Coast Campus, Griffith University,
Queensland 4222, Australia. †
These authors contributed equally to this work
*Correspondence and requests for materials should be addressed to Hua Gui Yang (
[email protected]) and Hai Feng Wang (
[email protected])
S1
Part I: Calculation section S1 Calculation details. All the spin-polarized calculations were performed with Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation using the VASP code, unless otherwise specified.1-2 The project-augmented wave (PAW) method was used to represent the core-valence electron interaction. The valence electronic states were expanded in plane wave basis sets with energy cutoff at 450 eV. The occupancy of the one-electron states was calculated using the Gaussian smearing (SIGMA=0.05 eV) for the solvent/surface interface models and Methfessol-Paxton smearing (SIGMA=0.01 eV) for clean surface. The ionic degrees of freedom were relaxed using the BFGS minimization scheme until the Hellman-Feynman forces on each ion were less than 0.05 eV/Å. The transition states were searched using a constrained optimization scheme,3 and were verified when (i) all forces on atoms vanish; and (ii) the total energy is a maximum along the reaction coordination but a minimum with respect to the rest of the degrees of freedom. The force threshold for the transition state search was 0.05 eV/Å. The dipole correction was performed throughout the calculations to take the polarization effect into account. As for model construction, considering that iodine molecule bond is up to 2.68 Å， a relative larger supercell is necessary to minimize the interaction between adjacent I atoms. The Pt (111) surface was modeled as a p(4×4) periodic slab with 3 atomic layers and the vacuum between slabs is ~20 Å (we did 4-layer pre-tests and got the same results with 3-layer calculations). The atoms in the bottom one layer are fixed, and all other atoms are fully relaxed. A corresponding 1×1×1 k-points mesh was used during optimizations after a 2×2×1 mesh pre-checking tests. Likely, a large four-layer p(4×4) Pt (100) and p(1×4) (411) surface were modeled with the bottom one layer fixed, corresponding a size of 11.26×11.26×24.97 Å3 and 11.94×11.26×35.91 Å3, and a 1×1×1 k-points mesh was applied to the two cases accordingly. The structures of I adsorption and the corresponding interface models are shown in Figure S1. S2
2.61Å
2.63Å
2.62Å
Figure S1. The structures of I adsorption at CH3CN/Pt(111) (left), CH3CN/Pt(100) (middle), and CH3CN/Pt(411) (right) interfaces, respectively. The corresponding bond length ( dPt-I ) is showed in red. To calculate the adsorption energy of surface species in the realistic solution, we take the effect of solvent implicitly into account by introducing several CH3CN solvent molecules into the surface model with a density of 0.79 g/cm3. The adsorption energy of I (EadI) is defined as: EadI=E(interface)+1/2E(I2)–E(I/interface) where E(interface), E(I2) and E(I/interface) are the energies of the liquid/electrode interface, I2 in the gas phase and I adsorbed on the liquid/electrode interface, respectively. Each individual energy term on the right side can be achieved directly from DFT calculation. The larger EadI is, the more strongly the species I bind on surface. It is worth noting that to take care of the configuration entropy contribution of CH3CN solvent and simulate the solution environment well, full optimization has been carried out for each configuration. For the subsequent calculations, the solvent molecules are all allowed to relax, aiming to achieve the corresponding stable solution environment in every step. S2
Effect of the coverage of the adsorbed solvent molecule and electrode
voltage on the material properties. Based on our calculation results, we found that solvent molecule CH3CN has a S3
relatively strong adsorption energy (0.42, 1.24, 1.21 eV on Pt(111), Pt(100), Pt(411) surface, respectively ) compared to EadI in gas phase (1.26, 1.43, 1.32 eV, accordingly). This means that the solvent molecule has appropriate adsorption coverage on the electrode surface. When I atom was introduced on the electrode surface, there should exist a competing adsorption effect between I atom and CH3CN molecule. We explored various co-adsorption configurations of I and CH3CN on Pt(111) surface by adjusting the number of CH3CN molecule adsorbed around I*. We found that it is the most favored thermodynamically when three CH3CN molecules adsorb nearby I*. As a result, this configuration was selected as the model to calculate the binding strength of I atom on the electrode surface, which gives an adsorption energy of 0.52 eV. Compared to the case of I adsorption in the absence of CH3CN solvent, it is clear that the solvent effect plays a negative role on I adsorption. To be systematic, we did a series of tests about the influence of solvent adsorption coverage on EadI. From Table S1, it is obvious that as the number of CH3CN molecules adsorbed around I*, the weakening on I adsorption is more evident, and the bond length (dI-Pt) of I-Pt Table S1. Influence of solvent adsorption coverage on EadI. Number of adsorbed CH3CN nearby I atom
Pt（111）
Pt（100）
Pt（411）
EaId /e
dI-Pt/Å
EadI /eV
dI-Pt/Å
EadI /eV
dI-Pt/Å
No solvent (gas phase)
1.26
2.56
1.43
2.55
1.32
2.55
0
0.93
2.56
\
\
\
\
1
0.65
2.57
1.57
2.58
1.42
2.58
2
0.59
2.59
1.53
2.59
1.38
2.62
3
0.52
2.63
1.56
2.61
4
0.42
2.68
\
\
increases accordingly. As for Pt(100) and (411) surface, we simulated the solvent environment with a higher coverage based on the following reasons: i) Our pre-tests show that these two low-coordinate surfaces have a much stronger adsorption capacity, S4
and ii) the surface coverage of solvent molecules has little effects on the adsorption energy of I atom (see Table S1). Considering that this is an electrode reaction, to better simulate the actual reaction condition, we took the electrode voltage effect into consideration. According to formula, U=Φ/e - USHE, it is clear that a certain electrode voltage (U) corresponds to a certain work function (Φ). To get a relative precise transition state barrier, we adjust the work function of transition state system equal to U+USHE by tuning the system net charge (take Pt(111) system for example, see Figure S2). For Pt(111) system, the experimental electrode voltage (U) is 0.61 V, and the value of USHE has been reported to be 4.43 V.4 So the work function should be 5.04 eV. As showed in Figure S1, for the neutral system, Φ is calculated to be 5.08 eV, a little bit larger than the experimental value (5.04 eV). After adding 0.1 electrons into this system, Φ is decreased to be 4.94 eV. The more electrons were added, a lower work function would be.
Figure S2. The electrostatic potential change along Z-axis on Pt(111) surface surrounded by CH3CN solvent. Using the above method, we calculated the desorption barrier of I atom at these S5
three interfaces under different electrode voltages. As shown in Figure S3, a linear relationship between the desorption barrier with the electrode voltage (U) with a small positive slope was achieved on all of these three interfaces. It shows that the desorption barrier goes higher as the electrode voltage increases, although it changes slightly. This is also consist with our thermodynamics results: ΔG0 (=2μΙ−(sol) - μI2(sol) 2μe, and we will show it below) is the Gibbs free energy change of half reaction -
-
I2(sol)+2e →2I (sol), and it is obviously the absolute value of ΔG0 becomes smaller as electrode voltage goes higher. With a decreasing thermal driving force, a corresponding decline in chemical activity is reasonable.
Figure S3. Desorption barrier under different voltage and the simulated linear relationship
S3
Free energy calculation.
Here we will show how to calculate the free energy change ( ΔG 0 ) of reaction (1). − I 2(sol) + 2e − → 2I(sol)
(1)
It is hard to directly calculate the energy of the charged periodic system accurately. However, we all know that the Gibbs free energy change of the standard hydrogen S6
electrode (SHE) reaction is zero ( ΔG H+ /H = 0 ), that is: 2
1 + + e − (SHE) H 2(gas) ↔ H (aq) 2
(2)
By combining reaction (1) and (2), we can get reaction (3), in which the eU term represents the electron free energy shift in the counter electrodes at the voltage U relative to the SHE. It is clear that the Gibbs free energy change ΔG1 of reaction (3) is equivalent to ΔG 0 , i.e. ΔG 0 = ΔG1 . − + I 2(sol) + H 2(gas) → 2I(sol) + 2H (aq) + 2eU
(3)
As shown below, we can design a thermodynamic cycle based on Hess’s Law to calculate ΔG1 indirectly:
(4)
−
+
I H For the above cycle, it is obvious that ΔG1 = ΔG 2 + 2(Εsol + Εsol ) − Δμ I2 . We used
Gaussian 03 software to calculate ΔG 2 in reaction cycle (4), and the solvation energies of I- in acetonitrile solvent, and H+ in water were taken from experimental −
+
I H , Εsol value( Εsol ), giving -2.86, -11.53 eV, respectively.5-6 For the value of Δμ I2 , that
is chemical potential difference of I2 molecule in between gas phase and CH3CN solvent, we have shown our calculation method in our previous work.7 S4
Determination of maximal ΔG0 (or the lowest electrode voltage)
ΔG0 is defined as the free energy difference of reaction I2+2e-Æ2I- (or I3-+2e-Æ3I-). It is easy to figure out that the lower the electrode voltage (U) is (in other word, the higher the electron energy level is), the larger the absolute value of ΔG0 is, as demonstrated in Figure S4. According to the principle of operation and energy level of DSCs, for the most common DSCs which use the TiO2 as the anode material, when S7
8 TiO the electrode voltage U = U CBM − U SHE ≈ -0.06 V (relative to SHE), ΔG0 would achieve 2
maximum.
S+/S* CBM
0 1
TiO2
eUmax
e-
I3-+2e-
hv
2
ΔG 0
3IVBM
Free Energy /eV
E vs SHE
/V
-1
S+/S
Anode
Cathode
Figure S4. Principle of operation and energy level scheme of the DSCs. Photo-excitation of the sensitizer (S) is followed by electron injection into the conduction band of TiO2. The dye molecule is regenerated by I-, which is regenerated at the counter electrode by electrons passed through the load. Potentials are referred to the standard hydrogen electrode (SHE).
Part II: Experimantal section S1 Synthesis of platinum nanocrystals with well-defined facets.
Pt nanocubes (Pt(100)) were prepared according to the literature.9-11 And the details are as follows: Under an argon atmosphere, Pt(acac)2 (20 mg, 0.05 mmol), oleyamine (8.0 mL), and oleic acid (2.0 mL) were loaded into a three-neck flask and heated to 135 ℃ under an argon stream. Tungsten hexacarbonyl (W(CO)6, 50 mg, 0.14 mmol) was then added rapidly into the vigorously stirred solution, and the S8
temperature was subsequently raised to 220 ℃ and kept for 40 min with vigorous with agitation. The resultant products were isolated by centrifugation and washed with anhydrous hexane and absolute ethanol for several cycles. The synthesis for Pt truncated nanooctahedrons (Pt(111)) was accorded to the literature9, but the carbon monoxide gas was replaced by W(CO)6. Under an argon atmosphere, Pt(acac)2 (20 mg, 0.05 mmol), oleyamine (8.0 mL), and diphenyl ether (2.0 mL) were mixed in the flash. Then the solution was then heated to 135 ℃ under vigorously stirring. W(CO)6 (50 mg) was added and the temperature was increased to 230 ℃ and kept for 40 min. The rest steps followed the synthetic procedures of Pt nanocubes. In a typical synthesis of Pt nanooctapods (Pt(411)),12 H2PtCl6 (20 mg/mL, 1 mL), poly(vinylpyrrolidone) (K30, 400 mg) and 0.3 mL methylamine solution (30%) were mixed together with 20 ml N, N-dimethylformamide. The resulting homogeneous yellow solution was transferred to a Teflon-lined stainless-steel autoclave with a capacity of 50 mL. Then, the autoclave was heated to 160 ℃ and kept for 10 h before it was cooled to room temperature. The products were separated via centrifugation and further washed with ethanol-acetone mixture for several times. S2
Preparation of dye-covered nanocrystalline-TiO2 electrodes.
To prepare the DSC working electrodes, the FTO glass was used as current collector (8 Ω/square, Nippon Sheet Glass, Japan). The commercially-available TiO2 powders, P25(Degussa, Germany), was used as raw material. A 12 μm thick layer (0.25 cm2) of P25 was loaded on FTO by screen printer technique.13 After sintering in S9
the air at 125 °C, a 4 μm thick scattering layer of 200 nm-sized TiO2 particles (HEPTACHROMA, DHS-NanoT200) was coated on the up layer. After sintering at 500 °C for 30 min, the TiO2-loaded FTO were immersed in a 5×10-4 M solution of N719 dye (Solaronix SA, Switzerland) in acetonitrile/tert-butyl alcohol (V/V=1/1) for 20-24 h to complete the sensitizer. S3 Preparation of Pt nanocrystals-counter electrodes.
To prepare the counter electrode, a hole was drilled in the FTO glass by sand blasting and pretreated according to the literature.13 The Pt nanocubes, nanooctahedrons, and nanooctapods pastes were fabricated as follows: 10 mg Pt nanocrystals were mixed with 0.8 g anhydrous terpinol and 1.12 g ethyl celluloses in ethanol (10 wt%), followed by stirring and sonication. The contents in dispersion were concentrated by evaporator, and the solid content was reached to 1 wt%. A single layer of Pt nanocrystals pastes was loaded on FTO by screen-printing procedure with a geometric area of about 0.36 cm2 and heated under airflow at 450 °C for 30 min. S4
DSCs assembling.
The Pt nanocrystals-counter electrode and dye-covered TiO2 electrode were assembled into a sandwich type cell and sealed with a hot-melt gasket of 25 μm (Surlyn 1702, DuPont). A drop of the electrolyte, which is prepared via dissolving 0.60 M 1-butyl-3-methylimidazolium iodide, 0.03 M I2, 0.50 M 4-tert-butyl pyridine, and 0.10 M guanidinium thiocyanate in acetonitrile, was put on the hole in the painted counter electrode. Then the electrolyte was introduced into the cell via vacuum S10
backfilling. The dummy cells for Tafel polarization measurements were assembled with two identical Pt nanocrystals loaded FTO, and the electrolyte was the same as the above. S5
X-ray diffraction (XRD) measurement.
XRD patterns were recorded on the dry membrane of aqueous dispersions of the three Pt nanocrystals on a glass wafer by a Bruker D8 Advanced Diffractometer (Japan) using Cu Kα radiation (λ=1.5406 Å) at 40 kV.
Figure S5. XRD patterns of the three samples of Pt nanocubes (black line), Pt
nanooctapods (red line) and Pt truncated nanooctahedrons (green line). The results are shown in Figure S5. By indexing these XRD patterns using standard ICDD PDF cards and reported data,9-10 it is confirmed that the as-synthesized three well-defined nanocrytals possess a face-centered cubic Pt phase with the Fm-3m space group. The sample of Pt nanocubes did not own a stronger (200) diffractions S11
than the (111) diffractions, which is due to that the typical bulk samples have the strongest (111) diffractions. According to the literature,9-11 when the nanocubes are carefully deposited on a surface-polished Si wafer, the (200) diffractions can be stronger than (111) diffractions. S6
Measurements of diffusion coefficient of the triiodide species.
Diffusion-limited currents within the dummy cells were determined by measurement of Tafel polarization using a slow scanning rate. Steady-state conditions were achieved. The diffusion-limited current density (Jlim) is proportional to the diffusion coefficient of I3-, D, according to equation (5) below, D=
l J lim 2nFC
(5)
where D is the diffusion coefficient of the I3-, l is the distance between the electrodes with the value of about 20 μm, C is the triiodide concentration of about 2.165×10-4 mol cm-3 and the rest retain their usual significance.14 And hence, some results for diffusion coefficient are summarized in Table S2. Diffusion coefficients can also be determined by measurement of the Nernst impedance in via transient-state test of EIS on DSCs, according to equation (6) below, W=
RT n F c0 A D 2
(6)
2
where R is the molar gas constant, T the absolute temperature, c0 the bulk concentration of I3-, A the electrode surface area, F the Faraday constant, n the number of electrons involved in the electrochemical reaction, D the diffusion coefficient of I3species. The diffusion coefficients of the samples obtained via the two different measurements are compared, and the results are listed in Table S2. S12
Table S2. Comparisons of the diffusion coefficients of the triiodide for different samples obtained from transient-state and steady-state measurements.
Samples
From transient-state measurements
Pt (111) Pt (411) Pt (100)
From steady-state measurements
W(sol) /Ω
D(I3-) / cm2 s-1
Jlim/ mA cm-2
D(I3-) / cm2 s-1
2.99 3.83 6.86
1.65 × 10-7 1.01 × 10-7 0.31 × 10-7
15.53 11.86 8.78
3.73 × 10-7 2.85 × 10-7 2.08 × 10-7
It can be seen that variation trends of the diffusion coefficient of the three samples from the two ways are in agreement, although the results in steady-state measurements are a bit larger, which may be ascribed to that the transfer of triiodide species in dummy cells is easier. S7
Measurements of electrochemical impedance spectra experiments under
different temperature Table S3 EIS parameters Rct and the calculated J0 of DSCs with different samples at
different temperatures Temperature /K 289 293 303 313
Pt (111) J0 / Rct / Ω mA cm-2 3.18 3.92 2.32 5.44 1.43 9.13 0.94 14.35
Pt (411) J0 / Rct / Ω mA cm-2 3.85 3.23 3.01 4.19 2.17 6.02 1.67 8.08
Pt (100) J0 / Rct / Ω mA cm-2 7.38 1.69 6.24 2.02 5.08 2.57 4.47 3.02
In order to further characterize the catalytic activity, EIS experiments were carried out at different temperature. The values of Rct were caculated from the EIS measurements, and the exchange current density (J0) were also obtained via equation (7) shown as below,
J0 =
RT nFRct
(7)
S13
where Rct is the kinetic component of the resistance determined by EIS data, and R, T, n and F have their usual significance. S8
Specific electrochemical activity of the three Pt nanocrystals electrodes
The electrochemically active Pt surface area was estimated from the charges involved in desportion of the underpotentially deposited hydrogen (HUPD) on the Pt nanocrystals surface via CV measurements.15-17 CV was conducted using a conventional three-electrode system in solution of 0.5M H2SO4. The Pt nanocrystals coated FTO were used as working electrodes as prepared in the experiment part of S2, a platinum foil as counter electrode and Hg/HgSO4/K2SO4(sat) reference electrode (MSE). The scanning range was set from -0.7 V to 0.8 V with a scanning rate of 50 mV s-1.
Figure S6. Cyclic voltammograms of Pt truncated nanooctahedrons, Pt nanooctapods and Pt nanocubes deposited on FTO in solution of 0.1M H2SO4 at a scanning rate of 50 mV s-1.
S14
To calculate the active surface areas of the catalysts, we calculated the charge transfer (QH) for the hydrogen adsorption and desorption of the catalysts from cyclic voltammograms (Figure S6). Assuming the charge per real cm2 of Pt with monolayer adsorption of hydrogen is Q0H = 210 μC/cm2,16 then the active specific surface areas can be obtained from Sact =
QH QH0
(8)
The calculation results are summarized in Table S4. Also, the actual masses used in each Pt CEs were measured by inductively coupled plasma-atomic emission spectra (ICP-AES, Thermo Electron Corp. Adv. ER/S). Before ICP measurements, the Pt CEs prepared via screen-printing procedure, which are identical to the CEs used in DSCs assembling and CV measurements, were immersed in chloroazotic acid with agitation for 24 h to thoroughly dissolve the Pt nanocrystals from the FTO substrate. The area-specific activity Rsct(Pt) which is obtained via Rct(Pt) (from EIS measurements) normalized by total electrochemically active surface area and the mass activity Rmct(Pt) obtained from Rct(Pt) normalized by Pt-based masses actually used in each sample have been calculated and compared. Table S4. Electrochemically active surface area (ECASA) estimated by measuring the charges associated with HUPD desorption in CV curves, Pt-based masses actually used in each sample and specific activities of different samples.
Samples
ECASA / m2 g-1
Pt (111) Pt (411) Pt (100)
10.48 14.92 14.29
total area / cm2 0.565 0.804 1.401
geometric area / cm2
Pt-based mass / μg
Rct(Pt) /Ω
Rsct(Pt) / Ω cm2
Rmct(Pt) / Ω μg
0.36 0.36 0.36
5.39 5.39 9.80
2.32 3.01 6.24
1.31 2.42 8.74
12.51 16.22 61.15
S15
References 1
Kresse, G. & Hafner, J. Ab initio molecular-dynamics simulation of the liquid-metal–amorphous-semiconductor transition in germanium. Physical Review B 49, 14251-14269 (1994).
2
Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set. Comput. Mater. Sci. 6, 15-50 (1996).
3
Alavi, A., Hu, P., Deutsch, T., Silvestrelli, P. L. & Hutter, J. CO Oxidation on Pt(111): An Ab Initio Density Functional Theory Study. Phys. Rev. Lett. 80, 3650-3653 (1998).
4
Reiss, H. & Heller, A. The absolute potential of the standard hydrogen electrode: a new estimate. The Journal of Physical Chemistry 89, 4207-4213 (1985).
5
Kelly, C. P., Cramer, C. J. & Truhlar, D. G. Single-Ion Solvation Free Energies and the Normal Hydrogen Electrode Potential in Methanol, Acetonitrile, and Dimethyl Sulfoxide. J. Phys. Chem. B 111, 408-422 (2006).
6
McIndoe, J. S., Tuck, D. e. n. n. i. s. G. & Crawford, E. The energetics of the X2 + X X3 equilibrium (X = Cl, Br, I) in aqueous and nonaqueous solution. Can. J. Chem. 84, 1607-1613 (2006).
7
Hou, Y. et al. Rational Screening Low-Cost Counter Electrodes for Dye-Sensitized Solar Cells. Nat.Commun. 4, 1583 (2013).
8
Kudo, A. & Miseki, Y. Heterogeneous photocatalyst materials for water S16
splitting. Chem. Soc. Rev. 38, 253-278 (2009). 9
Wu, J., Gross, A. & Yang, H. Shape and Composition-Controlled Platinum Alloy Nanocrystals Using Carbon Monoxide as Reducing Agent. Nano Lett. 11, 798-802 (2011).
10
Zhang, J. & Fang, J. A General Strategy for Preparation of Pt 3d-Transition Metal (Co, Fe, Ni) Nanocubes. J. Am. Chem. Soc. 131, 18543-18547 (2009).
11
Zhang, J., Yang, H., Fang, J. & Zou, S. Synthesis and Oxygen Reduction Activity of Shape-Controlled Pt3Ni Nanopolyhedra. Nano Lett. 10, 638-644 (2010).
12
Huang, X., Zhao, Z., Fan, J., Tan, Y. & Zheng, N. Amine-Assisted Synthesis of Concave Polyhedral Platinum Nanocrystals Having {411} High-Index Facets. J. Am. Chem. Soc. 133, 4718-4721 (2011).
13
Ito, S. et al. Fabrication of screen-printing pastes from TiO2 powders for dye-sensitised solar cells. Prog. Photovoltaics Res. Appl. 15, 603-612 (2007).
14
Wang, M. et al. CoS Supersedes Pt as Efficient Electrocatalyst for Triiodide Reduction in Dye-Sensitized Solar Cells. J. Am. Chem. Soc. 131, 15976-15977 (2009).
15
Doña Rodríguez, J. M., Herrera Melián, J. A. & Pérez Peña, J. Determination of the Real Surface Area of Pt Electrodes by Hydrogen Adsorption Using Cyclic Voltammetry. J. Chem. Educ. 77, 1195 (2000).
16
Xing,
Y.
Synthesis
Uniformly-Dispersed
and High
Electrochemical Loading
S17
Pt
Characterization
of
Nanoparticles
on
Sonochemically-Treated Carbon Nanotubes. J. Phys. Chem. B 108, 19255-19259 (2004). 17
Mayrhofer, K. J. J. et al. Measurement of oxygen reduction activities via the rotating disc electrode method: From Pt model surfaces to carbon-supported high surface area catalysts. Electrochim. Acta 53, 3181-3188 (2008).
S18
