Platinum oxide reduction kinetics on polycrystalline platinum electrodes

(1)

by

Geer Qile

B.Sc., Nankai University

A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science

in the Department of Chemistry

c

° Geer Qile, 2016 University of Victoria

(2)

Platinum Oxide Reduction Kinetics on Polycrystalline

Platinum Electrodes

by

Geer Qile

B.Sc., Nankai University Supervisory Committee

Dr. D. A. Harrington, Supervisor (Department of Chemistry)

(3)

Dr. D. A. Harrington, Supervisor (Department of Chemistry)

Dr. A. G. Brolo, Departmental Member (Department of Chemistry)

Abstract

A kinetic study on polycrystalline platinum (Pt) in sulphuric acid is presented. An electrochemical kinetic mechanism of Pt oxide reduction and surface oxide structures are proposed. The reduction reaction was studied by cyclic voltammetry (CV) and various potential programs that combine sweep and hold periods by an assembled analog instrumentation. The reduction peak was studied under three surface con-ditions: same oxide coverage  and same potential , diﬀerent  and same , and same  but diﬀerent , to determine the influence of  and  on the peak poten-tial p and peak shape. The double-layer charge measured previously by dynamic

electrochemical impedance spectroscopy (dEIS) was used to correct the CV base-line. Diﬀerential-equation-based models as a function of  and  were investigated to simulate the oxide reduction and oxidation, and estimate kinetic parameters. A simple mechanism combining desorption and multi-layer growth mechanisms showed good fit with both the spread-out oxidation peak and the sharp reduction peak. A microscopic surface oxide growth model was proposed to explain the surface oxides reduction mechanism.

(4)

List of Tables

5.1 Kinetic parameters to fit varying sweep rate data. . . 103 5.2 Kinetic parameters calculated from peak potential plots. . . 107

(8)

List of Figures

2.1 The voltage drop at the double layer. . . 6

2.2 Currents at the double layer. . . 7

2.3 An equivalent circuit of the electrode interface. . . 8

2.4 Coverage Example. . . 9

2.5 Cyclic Voltammetry for a polycrystalline Pt electrode in 0.5 M H2SO4 solution. . . 12

2.6 Diﬀerent-sweep-rate cyclic voltammogram for a polycrystalline Pt elec-trode in 0.5 M H2SO4 solution. . . 15

2.7 Capacitance at diﬀerent sweep rates. . . 16

2.8 Transition from reversible to irreversible behavior for oxide formation as a function of sweep rate by Langmuir kinetics. . . 18

2.9 Scheme of the place exchange and Pt oxide formation. . . 22

2.10 Microscopic model of surface oxides formation on Pt(111) - Pt buckling. 24 2.11 Microscopic model of surface oxides formation on Pt(111) - Pt buckling and O place exchange. . . 25

2.12 Microscopic model of surface oxides formation on Pt(111). . . 26

3.1 The equivalent circuit inside the electrochemical cell. . . 32

3.2 The input waveform diﬀerence between digital and analog potentiostats. 33 3.3 A simplified schematic of the Gamry potentiostat. . . 34

3.4 Sketch of the Analog instrumentation. . . 38

3.5 Schematic of the analog instrumentation setup. . . 39

3.6 The sweep-hold waveform. . . 42

3.7 Regular cyclic voltammograms by analog instrumentation. . . 43

3.8 Sweep-hold-sweep cyclic voltammograms by analog instrumentation. 43 3.9 Oxide charge density (Analog) from regular CVs (black) and SHS CV (red) vs sweep rate. . . 44

3.10 Oxide reduction peak potential comparison between regular CV (black) and SHS CV (red). . . 46

3.11 The step size influence on the hold potential. . . 47

3.12 Current response with time by Gamry Sequencer. . . 48

(9)

3.15 CV comparison between the Gamry Framework, VFP and analog

in-strumentation. . . 52

3.16 Oxide reduction charge density of regular CV by Gamry (blue) and analog instrumentation (black). . . 53

4.1 Cyclic Voltammetry at diﬀerent hold potentials and times. . . 56

4.2 SHS Cyclic Voltammetry by Analog instrumentation. . . 58

4.3 SHS CV at high sweep rate by Analog instrumentation. . . 58

4.4 The peak potential dependence on the sweep rate. . . 59

4.5 The peak current density dependence on the sweep rate. . . 59

4.6 Peak capacitance dependence on the sweep rate. . . 60

4.7 Cyclic Voltammograms of diﬀerent upper potentials. . . 61

4.8 Diﬀerent upper potential cyclic voltammogram of holding at 1.0 V for 180 s. . . 63

4.9 SHS Cyclic voltammograms of diﬀerent holding times. . . 64

4.10 Cyclic Voltammetry holding at diﬀerent cathodic potentials for 1 s. . 65

4.11 Cyclic Voltammetry holding at diﬀerent cathodic potentials for 3 min. 65 4.12 Holding at anodic potentials ( 1.2 V) for 3 min. . . 66

4.13 Cyclic voltammogram of same coverage at diﬀerent hold potentials. . 67

4.14 Cyclic voltammogram of same coverage at lower potentials. . . 68

4.15 Hold at anodic potentials for 3 min. Upper potential 1.4 V. . . 68

4.16 Hold anodic potentials for 3 min. Upper potential 1.3 V. . . 69

4.17 Cyclic voltammogram of same coverage and same potential. . . 70

4.18 Cyclic Voltammetry of the oxidation peak after holding at cathodic potentials. Hold time: 1 s. . . 71

4.19 Cyclic Voltammetry of the oxidation peak after holding at cathodic potentials. Hold time: 3 min. . . 71

4.20 Time eﬀect of the oxidation peak after holding at 0.9 V. . . 72

4.21 Double layer charge measurement. . . 73

4.22 Double layer charge at diﬀerent oxide coverages. . . 74

4.23 Double-layer capacitances versus sweep rates. . . 75

5.1 Cyclic voltammograms of holding at 0.8 V for diﬀerent times. . . 77

5.2 Oxidation and reduction charges. . . 78

5.3 Double layer current baseline correction. . . 80

5.4 Equivalent circuits for dEIS. . . 82

5.5 Double-layer charge baseline correction. . . 83

5.6 SHS CV pseudocapacitance diﬀerence with sweep rates. . . 85

5.7 Sweep rate dependence of reduction peak. . . 86

5.8 PtO formation diagram with two electrons transferred simultaneously. 87 5.9 PtO irreversible model (blue) with two simultaneous electrons trans-ferred. . . 88

(10)

5.10 PtO quasi-reversible model (blue) with two simultaneous electrons

transferred. . . 90

5.11 Pt2O formation diagram with two electrons transferred. . . 91

5.12 The relationship between clean Pt coverage  and electron coverage Θ 94 5.13 Pt2O multi-layer model (blue) with two simultaneous electrons trans-ferred. . . 96

5.14 Pt2O multi-layer model with one electrons transferred per step. . . . 100

5.15 Diﬀerent-upper-potential model fitting. . . 101

5.16 Occupied Pt coverage 1 −  (a) and electron coverage Θ (b) vs potential.102 5.17 Comparison of electron coverage Θ (red) with occupied Pt coverage 1_{−  (purple). . . 103}

5.18 Pt2O multi-layer model at diﬀerent sweep rates. . . 104

5.19 Experimental results of oxide reduction peak potential p vs log . . 106

5.20 Experimental results of full width half maximum vs log . . . 107

5.21 Pt2O multi-layer model (reduction) at the same potential with diﬀerent coverages. . . 108

5.22 Pt2O multi-layer model (oxidation) at the same potential with diﬀerent coverages. . . 109

5.23 Pt2O multi-layer model (oxidation) at diﬀerent coverages. . . 110

5.24 PtO unit cell. . . 113

5.25 PtO crystal structure, showing four unit cells. . . 114

5.26 −PtO2 unit cell. . . 115

5.27 −PtO2 crystal structure. . . 115

5.28 Possible surface-oxidation reaction mechanism drawn on Pt(111) surface.116 5.29 Possible surface-oxidation scheme on Pt(111) surface. . . 117

(11)

Nomenclature

Symbol Meaning Units

 Surface area cm2

 Capacitance F cm−2

dl Double layer capacitance F cm−2

f Faradaic capacitance F cm−2

p Peak capacitance F cm−2

 Potential V

hold Hold potential V

in Initial potential V

measured Measured potential V

out Potential output V

p Peak potential V

r Reference potential V

real Real potential V

up Upper potential V

 Faraday’s constant C mol−1

 Current A

out Current output A

 Current density A cm−2

(12)

f Faradaic current density A cm−2

p Peak current density A cm−2

 Electrochemical equilibrium constant various

 Rate constant various

eq _{Equilibrium rate constant} _various

r _{Rate constant (vs RHE)} _various

 Number of electrons 1

 Charge C

CPE Constant phase element parameter F cm−2 s−

 Gas constant J K−1 _mol−1

f Faradaic resistance Ω cm2 m Measurement resistor Ω cm2 s Solution resistance Ω cm2 u Uncompensated resistance Ω cm2  Temperature K  Time s f Faradaic impendence Ω cm2

(13)

 Symmetry factor 1

Γ Surface concentration mol cm−2

Γm Monolayer surface concentration mol cm−2

 Surface coverage 1

 Overpotential V

 Surface coverage 1

O Oxygen coverage 1

p Coverage at peak potential 1

Θ Electron coverage 1

Θmax Maximum electron coverage, 1

 Reaction rate mol cm2 _s−1

 Charge density C cm−2

m Monolayer charge density, equals 220 C cm-2 C cm-2

max Maximum charge density C cm-2

ox Oxidation charge density C cm-2

 Potential sweep rate V s−1

 Clean Pt coverage 1

(14)

Acknowledgements

I would like to thank my supervisor Dr. David Harrington for guiding me through the amazing maze of electrochemistry. With his professional guidance, my academic life in UVic has been a great achievement and enjoyment. I give special thanks to Dr. Frode Seland for his help and guidance in the instrumental setup. I also thank Thomas Holm for teaching me laboratory skills and Espen Fanavoll for his help with the VFP. Lastly, I thank the Natural Sciences and Engineering Research Council of Canada, the University of Victoria, and the Chemistry Departmental staﬀ for support of the research.

(15)

(16)

Introduction

The polymer electrolyte membrane fuel cell (PEMFC) is one of the renewable auto-motive energy technologies that produce electricity from electrochemical oxidation of hydrogen. But the cost, performance, and durability are still challenges for commer-cialization. The cathodic oxygen reduction reaction (ORR) is six or more orders of magnitude slower than the anodic hydrogen oxidation reaction (HOR), and the irre-versibility of the ORR restricts its eﬃciency and incurs about 40 % of all irreversible energy losses [1, 2].

Platinum is a widely-used catalyst for improving ORR activity in industry. How-ever, the dissolution of Pt degrades the catalyst and limits its lifetime. Even though Pt is only about 0.1 % of the fuel cell stack volume, it takes more than 25 % of the total cost [2]. Pt oxide is known to play a role in promoting the dissolution during the oxidation and reduction of Pt oxides [3] and is thought to degrade the performance of the ORR [4]. It is well established that an oxide film on the metal aﬀects anodic reactions in various ways, including blocking platinum adsorption sites [5], influencing intermediate adsorption, changing the electronic properties of the platinum surface, and aﬀecting the double layer [6]. Sugawara et al. [7] proposed that the Pt oxide does not alter the adsorption energy of reaction intermediates very much whereas

(17)

take place in a potential region in which the surface of platinum is partly covered by oxygenated species. Hence, the mechanism of the electrochemical formation and reduction of surface oxides is fundamental to understand how surface composition can determine the reactivity of the catalyst [4].

Even though studies of oxide formation and reduction at the Pt surface have been done for many years [8—12], the details of the oxide structure and the mechanism of oxide formation and reduction are still not fully resolved. In the case of the kinetics, most work has been directed at the oxide formation, and the reduction kinetics have been relatively neglected. A detailed understanding of the Pt oxide reduction mechanism is a crucial step for improving the ORR catalyst performance and durability and decreasing Pt loading. In this work, sweep-hold-sweep (SHS) cyclic voltammetry was used for a direct study on Pt oxide reduction.

In this thesis, Chapter 2 gives electrochemistry background knowledge, Chapter 3 introduces experimental methods used in this study, Chapter 4 shows experimental results on oxide reduction, Chapter 5 interprets experimental results and presents a kinetic model to simulate surface reactions, and Chapter 6 gives conclusions and proposes future work. In Chapter 2, the double layer concept, surface adsorption reaction kinetics - Langmuir kinetics, cyclic voltammograms on polycrystalline Pt electrodes, and literature reviews on Pt oxides are given. Even though Pt oxides have been studied for a long time, there are still diﬀerent schools of thought about reaction intermediates, oxidation states, and surface oxide structures. As for the double-layer current baseline, most studies still use the instrumental zero line or the double-layer region current to define the oxide coverage. In Chapter 3 the explanation of experi-mental methods used in this study is given. Of particular focus is sweep-hold-sweep (SHS) CV which was used to create identical surface conditions for oxide reduction. The simplest sweep-hold method involves preparing the surface with potential sweeps and then holding the potential at selected potentials. Constant potential is

(18)

typi-cally better for quantitative analysis of kinetics since the rate constants are fixed. Compared to conventional large potential steps, the sweep-hold preparation enables the initial conditions to be closely controlled, and reduces the double-layer charging. Adding different negative-going sweeps after identical sweep-hold experiments allows for study of sweep rate dependence under identical coverage and potential conditions, which is not the case for regular cyclic voltammetry at different sweep rates. We com-pared the difference between SHS by analog instrumentation and regular CV by a Gamry potentiostat, and showed the limitations of different SHS methods, the Gamry sequencer, Gamry Virtual Front Panel (VFP), and analog instrumentation.

Chapter 4 shows that the reduction peak potential p and peak shape depend

on oxide coverage  and potential . Three surface conditions: same oxide cover-age  and same potential , different  and same , and same  and different  were studied. By varying the preparation step to produce the same  at the same  but with different histories, evidence was found that the reduction peak position and shape only depend on oxide coverage  and potential , and not on the history. In addition to that, we look into how sweep rate affects reaction mechanisms, and use Tafel analysis to extract kinetic parameters, symmetry factor  and rate constant . In Chapter 5, we demonstrate how a differential-equation-based model simulates the complicated electrochemical reactions and how this model can be used to extract kinetic information. By building an electrochemical kinetic model, we can relate cur-rent peaks from CV to electrochemical reactions, and simulate intermediate reactions, kinetic parameters 1, −1, , , and surface oxide structures. A simple irreversible

desorption model was used to simulate the reduction peak. Then refinement was made by using the idea that the relationship between coverage and available Pt sites may not be linear [13]. A more complicated quasi-reversible model that combined multi-layer oxide growth with desorption gives reasonable agreement with both oxi-dation and reduction peaks. A more accurate double-layer charge baseline correction method was also proposed. A microscopic surface oxide growth model was proposed

(19)

thesis on the reduction mechanism on polycrystalline Pt and suggestions for further work are given in Chapter 6.

(20)

Chapter 2 Theories and literature review

2.1 Introduction

This chapter gives an introduction of electrochemical theories related to this thesis and the primary experimental tool: cyclic voltammetry. Simple mechanisms relevant to Pt oxide formation are given as examples, and the literature of Pt oxide formation and reduction is reviewed.

2.2 The electrical double layer

The interface between the electrode and the electrolyte is the main focus in electro-chemistry, and the interfacial region is called the "double layer". In the double-layer region, a voltage drop occurs between the electrode and the electrolyte over 3 Å to 10 Å as in Fig 2.1. This voltage drop ∆_m−s = m− s is the driving force for

electron-transfer reactions in which electrons jump across the interface from electrode to electrolyte (reduction) or from electrolyte to electrode (oxidation). As in Eq. (2.1) and Eq. (2.2), the key reactions produce electricity from electrochemical oxidation of

(21)

+



_m



_s

-working

electrode

electrolyte

Figure 2.1: The voltage drop at the double layer. _mis the potential of the electrode. _s is the potential of the electrolyte.

hydrogen and reduction of oxygen in the hydrogen fuel cell.

H2(g) À 2H++ 2e− (2.1)

4H++ O2(g) + 4e− À 2H2O (2.2)

For oxidation, as electrons are negative, so increasing ∆_m−s increases the ten-dency for the oxidation reaction to occur, and increases the rate of the oxidation. Conversely making ∆_m−s more negative increases the rate of the reduction reaction. This voltage drop is created when the electrode is in contact with the electrolyte, in less than a few microseconds. At any interface, even without obvious electron donors and acceptors, there is a voltage drop because impurities or solvent can react. The electrode potential is just ∆_m−splus a constant that depends on the reference elec-trode chosen. This interface may be simplified as a parallel-plate capacitor with two planes of equal but opposite charges, as in Fig. 2.2.

(22)

(re-working

electrode

electrolyte



_m



_s

+

-+

j

_dl

j

_f

Figure 2.2: Currents at the double layer.

action) current density f and the double-layer charging current density dl.

 = f+ dl (2.3)

The double layer current does not involve real charge moving across the interface. Even in the absence of electron transfer, the double layer current still flows as the changing of the voltage drop causes charging and discharging of the double layer. For oxidation, positive charges brought in from the wire constitute a positive current  and makes the charge density  in C cm−2 _{on the electrode more positive. The}

electrons jumping from the electrolyte to the electrode f, makes the charge on the

electrode more negative.



 = − f = dl (2.4)

The double layer can be modeled as a capacitor of which the double layer capacitance dl is approximately independent of potential.

dl= dl



 (2.5)

As the charges of this capacitor can be changed by an electron transfer reaction, the reaction is like a resistor in parallel with the double layer capacitor. Thus the

(23)

C_dl

j_f

R_f j

Figure 2.3: An equivalent circuit of the electrode interface.

equivalent circuit is a parallel arrangement of the double layer capacitance dl and

the Faradaic resistance f (Fig. 2.3).

2.3 Langmuir kinetics

The Langmuir isotherm is the simplest adsorption isotherm, which describes the rela-tionship between coverage  and the potential . The Langmuir isotherm assumes no interaction between adsorbed species, all adsorption sites have the same energy, the possibility of adsorption does not depend on whether the adjacent sites are occupied or not, and one site can only adsorb once.

For a surface reaction, in which the molecules adsorb on the electrode surface, such as Eq. (2.6),

Pt + H2OÀ PtO(ads)+2H++ 2e− (2.6)

we need to relate surface kinetics to surface concentrations. Γ is the symbol for surface concentration in mol cm−2, but usually coverage  is used. The definition of  is:

 = number of adsorbed atoms

(24)

Pt

O

Pt

O

(a)

(b)

Figure 2.4: Coverage Example. (a) O adsorption with one O atom per Pt atom. (b) one O atom per two Pt atoms.

The coverages of absorbed O atoms in Fig. 2.4 are (a)  = 4₅ ,and (b)  = 2₅, and the coverage of uncovered Pt surface atoms (available sites) is Pt = 1₅ for both (a) and

(b). The relationship between Γ and  is as Eq. (2.8) and Eq. (2.9):

Γ = Γm (2.8)

Γm =

m

 (2.9)

m is the charge density required to provide 1 e−/Pt, which we take to be 220 C

cm-2 _{at the polycrystalline Pt surface for adsorption of H as in Eq. (2.10),}

Pt + H++ e− À PtH(ads) (2.10)

where the notation of PtH(ads) implies one H per Pt.  is the Faraday constant,  = 96485 C mol−1. Γm is the surface concentration of the top layer of Pt atoms on

the clean surface, m represents 1 e−_/Pt.

Γm= m  = 220_{× 10}−6 _{C cm}-2 96485 C mol−1 = 228× 10 −9 _{mol cm}−2 _(2.11)

For a simple one-step reduction reaction Eq. (2.12):

PtO(ads)+2H++ 2e− À1

₋₁

Pt + H2O (2.12)

(25)

 = 1

£

H+¤2_{− }₋₁(1_{− )} (2.13)

1 is the cathodic (reduction) rate constant, and −1 is the anodic (oxidation) rate

constant.

The reaction rate  and the surface coverage  are related by Eq. (2.14), ΓPtO

 = Γm 

 =− (2.14)

we can find the coverage as a function of time, solve the diﬀerential equation and calculate the reaction rate .

The Faradaic current density f is then found by proportional to the

electron-transfer reaction rate  by Eq. (2.15):

f =



 =−2  (2.15)

The number two is the number of electrons transferred in Eq. (2.12).

A result from transition state theory allows the rate constants 1, −1 to be related

to the activation free energy and hence to the potential:

1 =  eq 1 exp µ −2 _ ¶ (2.16) ₋₁ = ₋₁eq exp µ 2(1_{− ) }  ¶ (2.17)

where the number two arises from the assumption of simultaneous two-electron trans-fer. The reference potential is chosen as the reversible potential r _{for the reaction,}

so the equilibrium potential  =  −r_{. }eq 1 and 

eq

−1 are rate constants at the

equilib-rium potential.  is the symmetry factor.  is the gas constant, 8314 J K−1 mol−1.  is the room temperature.

(26)

Combining Eq. (2.13), Eq. (2.16) and Eq. (2.17) gives the reaction rate as Eq. (2.18):  = eq₁ exp µ −2   ¶ [H+]2_{− }₋₁eq exp µ 2(1_{− ) }  ¶ (1_{− )} (2.18)

If we assume the H+ _{has fast mass transport, and so its concentration [H}+_] _at

the surface is constant, therefore can be subsumed into the rate constants. And we just write  in the exponential, meaning the experimental potential, as the reference electrode is the reversible hydrogen electrode (RHE = 0). Then the reaction rate and the Faradaic current density can be transformed into:

 = r₁exp µ −2 _ ¶ _{− }₋₁r exp µ 2(1_{− ) }  ¶ (1_{− )} (2.19) f = −2 µ r₁exp µ −2 _ ¶ _{− }r₋₁exp µ 2(1_{− ) }  ¶ (1_{− )} ¶ (2.20)

These equations and variations of them form the basis for the modelling and interpretation of the results.

2.4 Cyclic Voltammetry

Cyclic Voltammetry (CV) is a technique that monitors the current response to the potential change, as the potential is swept forward and backward at a linear rate. CV is the most sensitive direct electrochemical technique for examination of both the thermodynamic and kinetic aspects of electrochemical surface processes. Each particular adsorption or deposition of species on the surface will correspond to an

(27)

Figure 2.5: Cyclic Voltammetry for a polycrystalline Pt electrode in 0.5 M H2SO4

solution. Sweep rate 100 mV/s. Analog result.

individual peak in the CV profile [14]. Thus CV allows one to distinguish various reactions through the current peaks, e.g., oxide growth, under potential desorption of hydrogen (HUPD), oxygen evolution reaction (OER), hydrogen evolution reaction (HER). By integrating the current density  with respect to time,  =R2

1  , even

a small charge can be accurately measured by CV.

2.4.1 Cyclic Voltammetry on polycrystalline Pt

CV is a common experimental tool to study Pt surface reactions. The surface crystal-lographic orientation and the electrolyte will influence the characteristic CV shape. For a polycrystalline Pt electrode in sulfuric acid, the CV profile is stable at a par-ticular sweep rate  =  after a few potential cycles as in Fig. 2.5.

(28)

re-gion between 0.04 V (vs RHE, all the potentials used in this thesis are relative to the reversible hydrogen electrode (RHE)) and 0.35V, is the hydrogen underpotential desorption reaction (HUPD):

PtH(ads)À Pt + H++ e− (2.21)

The charge density m under the HUPD is assumed to be 220 C cm−2, and is used

to normalize the real electrode surface area  = ³R2

1 

´

m. As the current is

related to the surface area, the larger the surface area, the larger the current. To rule out the influence of diﬀerent electrode sizes, the current density  =  is plotted vs .

The two merged peaks at 0.12 V and 0.27 V are considered as desorption of weakly (0.12 V) and strongly (0.27 V) adsorbed hydrogen [15—17]. The region between 0.35 V and 0.8 V, is the double layer region with no Faradaic reactions, dl = dl_.

The double layer undergoes charging and discharging with the potential sweep, and forms a flat current region. However, the solution impurities, surface contamination, trace amount of O2, and heavy metal desorption may appear in this region [15].

The spread out peak between 0.8 V and 1.4 V is the region where adsorbed OH, chemisorbed oxygen, sub-surface oxygen, initial Pt oxide structures and surface-oxide phases coexist Eq. (2.22) and Eq. (2.23). Their existence strongly depends on the electrode potential and total surface coverage [17].

Pt + H2O À PtO(ads)+2H++ 2e− (2.22)

Pt + H2O À PtO+2H++ 2e− (2.23)

The notation PtO(ads) implies one O adsorbs on one Pt, and the notation PtO represents the phase formation with higher O coverage and more complicated

(29)

struc-the initial stage of struc-the oxide formation. PtO represents struc-the surface lattice Pt:O ratio rather than stoichiometric compound, as the thickness of oxide films doesn’t reach the "bulk" thickness [14]. According to [18], if the potential is held at 1.8 V, thicker and bulk-type oxide films will be formed. The oxide film has a maximum charge ca. 880 C cm−2, which corresponds to a maximum e− _{coverage, Θ}

max = 880 C cm−2220

C cm−2 = 4 e− _{per Pt. As the Pt oxidation state is not believed to exceed +2 [6],}

there must be multi-layer oxide growth.

When the positive-going sweep goes to higher than 1.55 V, the oxygen evolution reaction (OER) occurs:

2H2OÀ 4H++ O2(g) + 4e− (2.24)

In reverse, in the negative-going sweep, the sharp peak between 1.4 V and 0.5 V is the Pt oxide reduction peak, Eq. (2.12). Between 0.35 V and 0.04 V , there are peaks for underpotential hydrogen adsorptions, Eq. (2.10). When the negative-going sweep goes lower than 0.04 V, the hydrogen evolution reaction (HER) occurs:

2H++ 2e−À H2(g) (2.25)

Hydrogen adsorption (negative-going sweep) and desorption peaks (positive-going sweep) have the same peak potential and peak shape, which is the characteristic of reversible reactions. In contrast, the asymmetry between Pt oxide formation and reduction peaks means the oxide reaction is irreversible.

Some diﬃculty in obtaining all information of reactions from the CV profile arises because the peaks overlap. As in Fig. 2.5, two hydrogen peaks at 0.12 V and 0.27 V overlap. The flat oxidation peak from 0.8 V to 1.4 V, is thought to be diﬀerent peaks overlapping [14], as is the oxide reduction peak. Although it is widely accepted that

(30)

Figure 2.6: Diﬀerent-sweep-rate cyclic voltammogram for a polycrystalline Pt elec-trode in 0.5 M H2SO4 solution. Sweep rate 50 mV/s (black), 100 mV/s (red), 200

mV/s (purple), 300 mV/s (blue), 400 mV/s (pink). Arrow shows trend of increasing sweep rate. Gamry results.

the irreversible reaction is Pt oxide formation and reduction, the exact nature of the oxide and its growth mechanism are still uncertain. And this is the focus of the thesis: to study the Pt oxide surface reactions by manipulating the potential, especially by variation of the sweep rate.

2.4.2 Reaction reversibility

The dependence of the  vs  profile on the sweep rate  can be used to diagnose the reaction reversibility and mechanisms and derive kinetic parameters [14], which is also the focus of the thesis. By using diﬀerent sweep rates as in Fig. 2.6, the kinetic properties of the surface reaction can be obtained from the CV profile changing with the sweep rate .

(31)

Figure 2.7: Capacitance at diﬀerent sweep rates. Sweep rate 50 mV/s (black), 100 mV/s (red), 200 mV/s (purple), 300 mV/s (blue), 400 mV/s (pink). Arrow shows trend of increasing sweep rate.

peak potentials stay the same with , which is the characteristic of reversible reactions. The oxide reduction peak potentials pshift to more negative potentials with  which

is the characteristic of irreversible reactions.

In a surface reaction, the current density is proportional to the sweep rate through the capacitance,  = . The capacitance (Fig. 2.7) is the sum of the faradaic pseudocapacitance f = f and the double-layer capacitance dl[19]. The

capaci-tance is constant in the hydrogen desorption and adsorption region, and changes with the sweep rate in the oxide reduction region. The larger oxide reduction peak at the lower sweep rate, suggests more oxide grown at slower sweep rates. As the double-layer capacitance is almost constant with sweep rate, the diﬀerence corresponds to diﬀerent reduction mechanisms, from reversible, quasi-reversible to irreversible [20].

Why does the sweep rate influence the reversibility of the reaction? As the sweep rate is how fast the potential changes, and the potential change is the reaction driving force. If the potential changes too fast, the actual reaction can’t keep up with the

(32)

reaction driving force, then the reaction becomes more irreversible.

Transition behavior by Langmuir kinetics

Fig. 2.8 [20,21] describes the oxide reduction peak potential p, peak current density

p and Full width half maximum (FWHM) dependence on the sweep rate  for oxide

formation by Langmuir kinetics, and shows how CVs at diﬀerent sweep rates can be used to diagnose the reaction reversibility and derive kinetic parameters  and r

1.

FWHM is the potential diﬀerence at the half peak current density. These dependences on the sweep rate are used to diagnose the oxide reduction reversibility in Chapter 5.

Irreversible case Pt + H2O 1 À ₋₁ PtOH(ads) + H++ e− (2.26)

If ₋₁ = 0, the peak current density p can be found to be proportional to  as Eq.

(2.27) by solving diﬀerential equations. Then the peak potential p can be calculated

to be proportional to log10 as Eq. (2.28), and the coverage at the peak potential is

as Eq. (2.29) [22]. p = µ max e ¶  (2.27) p = ln 10  log10 +   ln µ max r 1 ¶ (2.28) p = 1− 1 e = 063 (2.29)

max= Γm, is the maximum charge density required to provide  e−/Pt. This

explains why the current density in the CV profile is proportional to sweep rate: to pass the same amount of charge in less time (at higher sweep rates), the rate must be larger, and so the current density is larger. For irreversible reactions, kinetic

(33)

E_p log reversible irreversible quasi-reversible log ₀ jp  reversible irreversible quasi-reversible log reversible irreversible quasi-reversible log ₀ F W H M

Figure 2.8: Transition from reversible to irreversible behavior for oxide formation as a function of sweep rate by Langmuir kinetics. Peak potential p, and peak width

(34)

parameters  and r

1 can be obtained from the slope and the intercept of p vs log 

plots of experimental results. These equations are combined with sweep-hold-sweep cyclic voltammetry results to derive kinetic parameters in Chapter 5.

Reversible case Suppose r1 and r−1 are both large, so we can assume r1(1− ) ≈

r

−1. Then we can again find out that the peak current density p is proportional to

the sweep rate . But the peak potential p does not change with the sweep rate .

p = µ max 4 ¶  (2.30) p = 05 (2.31)

No kinetic parameters can be extracted from data in the reversible case. In Fig. 2.6, the oxide reduction peak potential changes with the sweep rate, which means the oxide reaction is not reversible.

Quasi-reversible case There is a transition region between the reversible and irreversible regions, called the quasi-reversible region (both forward and backward terms are important). The transition sweep rate 0 (behavior changes from reversible

to irreversible), can be used to characterize the reaction rate and calculate kinetic parameters. The quasi-reversible case is the focus of the modelling, and is discussed in detailed in Chapter 5.

Even though there have been studies on the oxide formation at diﬀerent sweep rates [20—22], a direct study on the oxide reduction is still absent. So a detailed kinetic study on the oxide reduction dependence on the sweep rate is studied here by sweep-hold-sweep cyclic voltammetry. The experimental details are discussed in Chapter 3. The reaction reversibility, mechanisms and kinetic parameters are discussed in Chapter 5.

(35)

Single crystalline surfaces, Pt(111), Pt(100), Pt(110), etc. are studied mostly as idealized, high-symmetry surfaces that can simplify the analysis of results. Because the polycrystalline electrode consists a mix of surface crystallographic orientations, Pt(111), Pt(100), Pt(110), reactions on polycrystalline Pt electrode may be related to reactions at the single-crystal electrodes. A model of polycrystalline Pt oxidation and reduction processes should be related to a description at single-crystal surfaces. Pt(111) is thought to be the dominant surface orientation in catalyst particles [18], and it is well-characterized and has high thermodynamic stability [23], and therefore mechanisms based on Pt(111) are widely discussed.

As for techniques, cyclic voltammetry (CV) is the primary tool to study oxide for-mation and reduction at Pt catalysts. Secondary experimental methods such as elec-trochemical quartz crystal micro-balance (EQCM) [24], Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS) [25], scanning tunneling microscopy (STM) [6], and theoretical methods such as density functional theory (DFT) [26], and lattice gas approaches [27], can be used as well to probe oxide formation and reduc-tion. Because of the necessity for the electrolyte to be in contact with the electrode, many of the high-vacuum surface science techniques are inapplicable [14]. Thus the direct observation of the surface structure by scanning tunnelling microscopy (STM), atomic force microscopy (AFM), and direct surface X-ray diﬀraction, are used widely for single crystalline electrodes.

2.5.1 Intermediates

The oxide growth mechanism on platinum in acidic solutions has been studied since the mid-1900s [28—30]. However, there are still disagreements on the intermediates in the surface reactions. Current experimental techniques having diﬃculty in detecting oxygen and hydrogen-containing species in the presence of electrolyte, especially at

(36)

low coverages [16]. Also it is diﬃcult to confirm the nature of the initial oxide species, because of the short time scale of platinum oxidation. OH(ads), chemisorbed O, subsurface O, initial Pt oxides, surface-oxide phases, H2O and adsorbed anions were

proposed in diﬀerent mechanisms. As an oxide on a metal surface should have at least two layers [14], and most surface oxide species generated on the electrode surface are very thin films are not bulk 3D materials, the notation of "oxide" is not well defined. Early studies assumed that a hydroxide species was deposited during the initial stages of growth because the anodic charge of the cyclic voltammogram was consistent with those structures [19]. The following mechanism was proposed [16, 31]:

Pt + H2O À Pt − OH(ads) + H +

+ e− (2.32)

Pt_{− OH(ads) → OH − Pt(place exchange)} (2.33)

PtOH _{→ PtO + H}++ e− (2.34)

The OH(ads) intermediate was widely used in kinetic models [2,31—33]. However, there is no direct evidence of the existence of PtOH by in situ electrochemical quartz crystal nanobalance (EQCN) [24], and it is thought to only appear in the initial state of the oxidation (≤ 092 V [34]). This might because OH(ads) is formed quickly and then removed or the coverage is small and it is not detected.

Jerkiewicz et al. [6] proposed that the chemisorbed species was oxide rather than hydroxide, identified the species as PtO from the molecular weight (15.8 g mol−1_{) by}

EQCN. This value was close to 16 g mol−1 for adding one ML O to Pt. They proposed a theoretical model on Pt(100) as in Fig. 2.9, and proposed the intermediates were chemisorbed O at 0.85 V ≤ E ≤ 1.10 V, and Pt2+−O2− at 1.20 V ≤ E ≤ 1.40 V .in

(37)

H2O molecule Pt surf ace atom O atom H+ _H+ e -e -Oads OH2

Figure 2.9: Scheme of the place exchange and Pt oxide formation. Eq. (2.35).

Pt + H2O → Pt − OH2 (2.35)

(Pt_{− Pt) − H}₂O _{→ (Pt − Pt) − O(ads)+2H}++ 2e− (2.36) Pt_{− O(ads) + H}₂O _{→ Pt}2+_−O2−+2H++ 2e− (2.37)

As O species were detected on the surface, O intermediate was widely accepted [35—38].

2.5.2 Place exchange

In early studies on Pt(111), ex situ low-energy electron diﬀraction (LEED) suggested the surface undergoes a surface reconstruction after cycling, and becomes roughened, with the observation of vacancies in the top layer of Pt [39]. A classical mechanism, the "place-exchange" theory was used to explain the reconstruction: metal and oxygen atoms switched positions at high oxygen coverage [14]. The place exchange is believed

(38)

to lead to an irreversible change of the surface. After place-exchange, either the exchanged Pt atoms are too far away from original sites, or irreversible oxide clusters are formed. This causes the structural defects and irreversible oxides at higher O coverage. However at low coverage of exchanged sites, the surface can still recover to its original state upon reduction.

AES data by Jerkiewicz et al. [6] demonstrated that the place-exchange process occurs in the 1.1—1.2 V potential range, as evidenced by the intensity of the O signal increasing with the increase of up (upper potential). Upon exposure to the electron

beam, the oxygen signal at 1.1 V ≤ up ≤ 1.2 V had a transitional behavior: it

disappears but more slowly than at 0.85 V ≤ up ≤ 1.1 V, which means the

O-containing species do not desorb easily, indicating an place-exchange. Teliska et al. [37] also found that the place exchange is not directly observed before 1.1 V by probing changes in the local atomic structure by in situ X-ray Absorption Spectroscopy. They applied the ∆ technique to X-ray absorption near edge structure (XANES) spectra and were able to assign diﬀerent O adsorption sites. At low coverage, OH adsorbs primarily in the atop sites on steps and edges of small Pt clusters. As the coverage increases O is found in bridged/fcc sites, first at the bridged site at edges, and then in the fcc sites on the faces. After a Pt-O(H) “place exchange” or oxide formation, O is found in subsurface sites, but this is not apparent in the ∆ spectral shapes until 1.1 V. Thus, the place-exchange is thought to occur above 1.05 V (vs RHE) after which a “surface oxide” forms.

You et al. [40] studied Pt(111) in 0.1 M HCIO4by X-ray reflectivity measurements,

and proposed the place-exchange is irreversible for charge transfers ≥ 1.7 e−_{/Pt or}

above 1.25 V (vs RHE), because of the decreased intensity of reflectivity at (00L) rods. However, for charge transfers ≤ 1.7 e−_{/Pt or below 1.25 V (vs RHE), the}

place-exchange is reversible, as the intensity returned to the clean surface data. The microscopic view of the place-exchange theory is that, when the oxygen cov-erage is above 0.5 ML, oxygen atoms shift positions to Pt subsurface, or Pt atoms

(39)

Figure 2.10: Microscopic model of surface oxides formation on Pt(111) - Pt buckling. Top view (left) and side view (right).

escape from the metal into the oxide. This process decreases the interaction energy between new adsorbed surface oxygen and surface oxides, and leads to irreversible ox-ide formation. However, there have been disagreements on the nature of the surface species: O may penetrate into the Pt lattice, or Pt may buckle into an O layer.

2.5.3 Surface structures

Devarajan et al. [41] observed the coexistence of p(2 × 1) and p(2 × 2) oxygen ad-sorption layers at 0.5 ML coverage by STM for the O2 interaction with Pt(111) under

ultrahigh vacuum (UHV) conditions, and proposed a structural model as in Fig. 2.10. They suggested a p(2 × 1) oxygen layer (gas phase) at 0.5 ML and one-dimensional (1D) Pt-oxide chains at 0.75 ML, and concluded that the initial formation at Pt(111) surface is the Pt oxide chain instead of chemisorbed phases or subsurface sites. Exper-imental STM images showed that at coverages approaching 0.75 ML, the chains form an interconnected network of Y-shaped structures with regions locally resembling a honeycomb. This suggests the place exchange appears above 0.75 ML. Hawkins et al. [42] also proposed two structural models at the coverage of 1 ML by density func-tional theory (DFT) based on STM images. In the two models by Devarajan et al. and Hawkins et al., Pt surface atoms buckled into the oxygen layer, and formed a PtO2 structure.

(40)

Pt buckling subsurf ace O

place exchange

Figure 2.11: Microscopic model of surface oxides formation on Pt(111) - Pt buckling and O place exchange. Side view.

Holby et al. [43] proposed a stable hybrid structure at 1 ML with both Pt buckling and O place-exchange as in Fig. 2.11. Their suggested surface model consists of two kinds of PtO2 stripes, and explains how oxygen can move subsurface by an almost

perpendicular rotation of the Pt buckling chain. Drnec et al. [44] proposed that the place-exchange Pt atoms are directly above their original positions in Pt(111), and the surface reconstruction begins at 1.02 V. They also found that O2in the electrolyte

does not influence surface oxidation significantly.

Imai et al. [33] studied Pt(111) nanoparticles via in situ and time-resolved X-ray diﬀraction measurements, and found an −PtO2 2D layer around 1.2 V, and a

quasi-3D −PtO2at 1.4 V. By analyzing the Pt-O bonds, they confirmed Pt-O bonds

lengths changed from 2.3 Å for Pt-OHH, 2.2 Å for Pt-OH, 2.0 Å for adsorbed atomic oxygen or Pt-O oxides, to 3.1 Å Pt-O-Pt bonds [45—49]. They proposed a microscopic model for Pt(111) oxide growth as in Fig. 2.12. They assumed 3-fold hollow sites (fcc or hcp) are more stable than on top adsorption for O. By an adsorption model, they explained the place-exchange in a microscopic way. The OH species adsorbed on top of Pt(111) top layer, then OH were oxidized to atomic oxygen, and moved to 3-fold hollow sites, and these two steps were reversible. In the next step, oxygen entered into the Pt lattice by place exchange, and formed a 2D layer. At higher potentials, −PtO2 formed by adding oxygen atoms to hcp sites.

Whether O penetrates into the Pt lattice, or Pt buckles into an O layer, or whether Pt atoms are directly above their original positions, or shift, still remains uncertain.

(41)

-PtO₂ -PtO₂

Figure 2.12: Microscopic model of surface oxides formation on Pt(111) by Imai et al. [33]. Side view.

Even though Pt(111) is thought to be a dominant facet in polycrystalline Pt and Pt nanoparticles, Conway et al. [18] concluded that the surface oxide formation kinet-ics depend on diﬀerent surface structures, and so the mechanisms may change in a polycrystalline Pt electrode where surface roughness and defects exist.

2.5.4 Modelling

Many researchers have attempted to simulate CVs to gain a better understanding of the interplay between oxide growth and platinum dissolution. However, only a small number of kinetic models are based on fitting polycrystalline Pt experimental CVs [2,14,28,31,36,50,51]. Each study is able to replicate some experimental behavior, but none have achieved the correct shape and behavior of CVs over a range of conditions. For instance, Tilak et al. [31] developed a polycrystalline Pt oxidation model, with a multi-step oxidation from water to hydroxide and oxygen. However the model was only consistent with the anodic features of the CV but was unable to provide a consistent picture of the voltammetry behaviour through the entire potential cy-cle. Appleby’s [28] work incorporated a two-step oxide reduction mechanism with a

(42)

pre-equilibrated step before the rate determining step. Modified Butler—Volmer ki-netics were implemented, where Temkin adsorption isotherms were applied to both the anodic and cathodic terms of the reaction rate. This form of the rate equation simulated the correct width of the reduction peak.

Heyd et al. [30] proposed the oxide growth law for the anodic oxide on polycrys-talline Pt.

ox

 = oexp () exp(−ox) (2.38)

The Temkin term exp(−ox) accounts for the energy of adsorption increasing with

coverage of oxygen. The current was predicted to rise to a plateau current whose value is proportional to the sweep rate, as observed experimentally. And the anodic current-potential relationship for cyclic voltammetry was predicted. Harrington [35] presented a microscopic model of the growth of the thin anodic oxide film on Pt. The model predicted correctly the observed direct logarithmic growth law for potential step transients, and explained many structural and kinetic aspects of Pt oxide growth. However, the slope was not in exact agreement with experiment.

Darling and Meyers [36] incorporated a two-step oxide (PtO) formation mechanism to describe the formation and reduction of platinum oxide. Modified Butler—Volmer kinetics based on Harrington [35], Heyd [30], Conway [29] and co-workers were imple-mented, where Temkin adsorption isotherms were applied to the anodic portion of the rate equation only, and its inclusion allowed the CV to reach a plateau at higher up.

However the model at higher potentials (1.0 V-1.4 V) only predicted the reduction peak potential and not the peak shape.

Redmond et al. [32] proposed a two-step oxidation, based on the EXAFS results. Place exchange on edge sites was modeled as a separate step. Place-exchanged oxides of varying energy states (from PtOH to PtO2) are formed through a single transition

(43)

al. [2] extended the kinetic model from Pt(111) to an initial analysis of polycrystalline Pt CVs within the potential range of 0.65—1.15 V in 0.5 M HClO4, and got a reasonable

fit. The model is based on Gomez-Marin’s model [16], and consists of diﬀerent stages of oxide formation from water to hydroxide and oxygen. However these models were only fitted within the potential range 0.65—1.2 V.

Dahlstrøm et al. [13] studied bulk CO oxidation on polycrystalline platinum in CO-saturated 0.5 M sulfuric acid. The goal was to build a model that reproduces the experimental current—potential characteristics, with special attention to the platinum oxide region. The reaction rate was modelled with diﬀerential equations based on the platinum oxide growth equation developed by Heyd and Harrington [30]. The model distinguished the coverage based on counting electrons, Θ, from the coverage based on counting occupied Pt sites, PtO, to allow for multilayer growth before a monolayer

of sites is occupied.

PtO =

1

2Θ Θ≤ 15 (2.39)

PtO = 1− 502exp(−2Θ) Θ ≥ 15 (2.40)

The model built around this empirical equation describing the rate of platinum ox-ide formation and reduction, produced good fits to the experimental results. Our proposed multi-layer model is based on this model.

Various surface species have been suggested, including Pt-OH(ads) or O(ads), PtO [6] or higher oxidized oxides [32,52], multiple oxides or single species with differ-ent oridiffer-entations, but there is little agreemdiffer-ent. A kinetic model that can fit the differ-entire CV profile of polycrystalline Pt is needed to help understand the reaction mechanism, and decide on likely surface species.

(44)

2.6 Double-layer charge correction

To simplify calculations, the instrumental zero line was used widely as the baseline to separate the oxidation charge and the reduction charge, inside of defining a double-layer charge baseline. However, the measured current density  is the sum of the Faradaic current density f and the double-layer current density dl, so it is important

to separate faradaic current from the double layer charging current by double-layer charge correction, to get accurate kinetic parameters.

The simplest double-layer charge correction is horizontal extrapolation of the CV response over the flat double-layer charging region at Pt (0.35 V to 0.85 V) into the oxide formation region [15, 53]. This method assumes the double layer capacitance is constant. However in the real surface, the double layer capacitance is thought to depend on the potential. Pell et al. [54] proposed that the interfacial double-layer capacitance at oxidized Pt surfaces is diﬀerent with the capacitance of the oxide-free Pt surfaces. The double-layer capacitance decreased from 80 F cm−2 _{to 40}

F cm−2 when the potential increased from 0.9 V to 1.4 V (vs RHE). The double-layer capacitance was separated by electrochemical impedance spectroscopy (EIS) in 0.5 M H2SO4 with 0-0.01 M CI−. At the initial stages, electrosorbed O and anions

adsorption were measured and coupled together with the double-layer capacitance, called pseudocapacitance. The decrease was thought to be caused by the desorption of anions, e.g. HSO−₄ because of the increase of electrosorbed O, or Pt2+_/Pt4+ _redox

couple. Conway et.al. [55] measured the change of double-layer capacitance behaviour of polycrystalline Pt electrodes by equivalent circuits. The double-layer capacitance dl decreased from 80 F cm−2to 40 F cm−2from 0.9 V to 1.4 V (vs RHE). The total

value increased from around 75 to 300 F cm−2 _{than decreased to 120 F cm}−2 _from

0.6 V to 1.0 V. Sacci [56] measured the double layer capacitance on polycrystalline Pt electrode by fitting the impedance (EIS) results to equivalent circuits. He suggested that the measured double-layer capacitance depends on the oxide coverage. At 50

(45)

the potential range of 0.7 V to 0.9 V, then decreased to 60 F cm−2 _{at 1.5 V in}

the positive-going sweep. The capacitance stayed almost constant at 60 F cm−2

from 1.4 V to 0.8 V in the negative-going sweep, then decreased to 35 F cm−2 _at

0.6 V. The double layer capacity of the Pt(100) electrode in slightly acidic KClO4

solutions [57] was determined by impedance measurements as function of potential. It was found to be between 20 − 30 F cm−2 _{except around -0.1 V vs Saturated}

Calomel Electrode (SCE), the capacity—potential curve exhibits a peak of about 50 F cm−2_{. Pajkossy et al. [58] also got similar double layer capacitance of the Pt(111)}

electrode by impedance and capacitance measurements in neutral and acidic aqueous perchlorate solutions.

In all, the double layer capacitance depends on the oxide coverage or the potential. To get a more accurate faradaic charge from CV, the subtraction of the double layer current is important. This thesis gave a direct experimental study on platinum oxide reduction on polycrystalline Pt, and proposed a simple desorption and multi-layer growth mechanism with the double-layer baseline correction.

(46)

Chapter 3 Techniques

3.1 Introduction

This chapter discusses the instrumentation, and in particular how sweep-hold-sweep (SHS) cyclic voltammetry was implemented by three diﬀerent methods: Gamry Se-quencer, Gamry Virtual Front Panel (VFP), and analog instrumentation. The ad-vantage of SHS CV and the diﬀerences and limitation of these three methods are discussed. The cell and other experimental details are also included.

3.2 Electrochemical cell

A three-electrode electrochemical cell contains the Working electrode (WE), the Ref-erence electrode (RE), and the Counter electrode (CE). The potential applied to the WE is controlled. The RE has a constant potential and is used to measure the WE potential. No current flows through the RE. The WE is the electrode where a surface reaction of interest takes place and the current is measured. The current passing through the WE, flows into (or out of) the solution, and then the CE completes the cell circuit.

(47)

WE CE R_f C_dl R_s R I I solution

Figure 3.1: The equivalent circuit inside the electrochemical cell. The equivalent circuit inside the cell is as in Fig. 3.1.

The double layer acts as a capacitor dl, f represents the Faradaic resistance

(reaction resistance). f and dl represent the interface at the WE surface in the

scale of 3-10 Å. s is the solution resistance between the tip of the RE and the WE

interface in few mm, also called the uncompensated resistance u. As the potential

applied to the WE is measured between the WE and the RE. s will result in a

potential drop s for the real potential across the interface.

real= measured− s (3.1)

To minimize the influence of the s, one can move the RE to make the tip closer

to the WE, or increase the conductivity of the solution (decrease the resistance). A potentiostat can supply a voltage to the CE, to get the desired voltage of the WE relative to the RE.

3.3 Potentiostats

Diﬀerent methods were investigated to generate the SHS waveform and measure the current response, including the Gamry Framework Sequencer (digital), the Gamry

(48)

E

t E

t _t

Original _Digital Analog

E

Figure 3.2: The input waveform diﬀerence between digital and analog potentiostats. Virtual Front Panel (digital) software and analog instrumentation.

3.3.1 Digital vs Analog

Analog and digital potentiostats are used to transmit the input signal to the elec-trochemical cell. The analog potentiostat applies the input waveforms as they are in a continuous way. However, a digital potentiostat samples analog waveforms into discrete numbers, and applies discontinuous voltage-time signals to the cell.

3.3.2 Gamry

The Gamry Reference 600 used in this work is a digital potentiostat. It generates the digital input signal with potential steps and applies this waveform to the cell. The Gamry controls the potential diﬀerence between the WE and the RE, and measures the current flow between the WE and the CE. A simplified schematic of the Gamry potentiostat is shown in Fig. 3.3.

The electrometer measures the potential diﬀerence between the RE and the WE, which is recorded as the Voltage output. The Control Amplifier is used to force the potential applied to the WE to stay the same as the signal, by adjusting the CE potential. The CE potential is adjusted by minimizing the error signal, which is the

(49)

Cell WE CE RE R_m I/E Converter Electrometer Signal ₊ -Control Amp x1 x1 Voltage Current*Rm CA Speed IE Stability

Figure 3.3: A simplified schematic of the Gamry potentiostat. Repro-duced from http://www.gamry.com/application-notes/instrumentation/potentiostat-fundamentals/ with permission.

(50)

diﬀerence between the output potential from the electrometer and the desired Signal potential. The Current-to-Voltage (I/E) converter measures the cell current. The Control Amplifier forces the current from the circuit to flow through a measurement resistor (m). The cell current flows to ground through m thus the voltage drop

across this is proportional to the current.

Gamry device settings

Gamry has internal filters (I/E stability and CA speed) to reduce the noise of signals.

I/E Range The I/E Converter measures the voltage drop across a resistor (m) to

calculate the current passes the cell. There are an array of diﬀerent resistors across several orders of magnitude. By changing the resistance, we can set the current range to a range large enough but still close to a given current. More sensitive ranges require larger resistors. By using an appropriate I/E Range at diﬀerent sweep rates, we can measure accurate and noiseless currents.

I/E Stability I/E Stability sets the I/E converter stability. There are four settings for stability, Fast (0), Medium Fast (1), Normal (2), and Slow (3). The Fast value corresponds to no extra I/E converter filtering. The Normal value corresponds to a small I/E filter. The Slow value corresponds to a large I/E filter which removes 50/60 Hz noise in the lower current ranges.

CA Speed The Control Amplifier (CA) controls and adjusts the signal that is applied to a cell. CA Speed sets the Control Amplifier stability. There are also four settings, Fast (0), Medium Fast (1), Normal (2), and Slow (3). Faster speed settings allow the control of fast signal changes. The fast value enables the CA to accurately apply fast changing currents to the cell, but the applied current may have power

(51)

may degrade the accurate display of the input signal.

Gamry Software

Gamry Framework Gamry Framework is software used to control the Gamry Po-tentiostat, set up waveforms and internal filters and record current response directly. The advantage is that the waveform is easy to set up, and the device settings (I/E Stability, CA Speed, etc.) are also set automatically by the Framework software ac-cording to diﬀerent current measurements. For the step size, there is a maximum point limit for a Gamry Framework experiment. The number of points is equal to the Scan Range (mV) divided by the Step Size (mV), which has a maximum of 64000 points for all cycles. If the scan range is 3 V, the minimum step size is 0.1 mV for two cycles. When sweeping to a higher upper potential up (larger scan range), an

increase of the step size is required to stay within the maximum points limit.

Virtual Front Panel Virtual Front Panel (VFP) is software used to control the Gamry Potentiostat, and it is an alternative to the Gamry Framework software. VFP works like Framework. It can be used to set the potential waveform, sweep rate, potentiostat settings, and record the current response. We can set up the signal by inputting a signal file, which can be generated by Excel or Matlab. The signal file contains one number per text line. Each number represents a potential, and the diﬀerence between two text lines is the step size. The step size can be set as small as you want, as long as the file size is within the maximum size. The step size is set diﬀerently according to the sweep rate and the Acquisition Frequency (AF). The sweep rate is controlled by the AF. For example, if the AF is set to 1000 Hz, 1000 lines will be read from the file each second. Then the sweep rate can be calculated,  = (1000− 1)Vs.

(52)

3.3.3 Analog potentiostat

General description

The Gamry potentiostat can also function as an analog potentiostat after running a short software script, which sets m. Then the potential is controlled from an

external function generator. The function generator can be used to generate analog waveforms. By applying an analog waveform to Gamry (analog mode), waveform is applied to the cell and the current is measured. Then an Oscilloscope is used to measure the analog potential and current response. In the experiments with analog instrumentation, function generators were used to generate the analog signal and the Gamry was used only to apply the signal to the cell. By applying an analog signal, artifacts due to step size are ruled out, and the signal is continuous. This instrumentation also allows for more complicated waveforms. As a result, we can accurately study surface reactions under diﬀerent surface conditions.

Setup

1) A Function Generator HB-111 (Hokuto Denko Ltd.) produces an analog sweep waveform for cyclic voltammetry. 2) An Agilent 33220A 20 MHz Function/Arbitrary Waveform Generator applies the hold-potential waveform. 3) A Princeton Applied Research Model 175 produces the analog sweep waveform after holding. 4) Stanford Research System (SRS) Small Instrumentation Modules (SIM) are used to combine the three waveforms. 5) A Gamry Reference 600 potentiostat is controlled with the Gamry Framework software, but is otherwise used as an analog potentiostat by applying a Script. The Script sets up the potentiostat device settings. 6) A Pico Technology Limited ADC-212 Virtual Instrument is used to record output potentials and currents with PicoScope software. A sketch of the analog instrumentation is shown in Fig. 3.4.

(53)

HB Function Generator

Agilent Function

Generator PAR Sweep _Generator

Eout Eout EXT TRIG IN EXT TRIG IN EXT SIG IN WE RE CE Eout Iout Cell PicoScope B A Computer Summing AMP Scaling Amp -1.00 Scaling AMP -1.00 TRIG SRS SIM900 Gamry

(54)

Ext Sig In Output Ext Sig In Output _Output

+

E_out Ext Sig In _WE RE CE

I

_out

A

B

HB

Agilent

PAR

Gamry

Cell

Picoscope

Computer

Sum Amp Scaling Amp

+

Figure 3.5: Schematic of the analog instrumentation setup. A detailed scheme of the instrumentation setup is as Fig. 3.5.

The HB-111 Function Generator generates the sweep waveform from the initial potential in (here we used 0.04 V) to the upper potential up (here we used 1.4 V),

then to the hold potential (hold). The continuous mode is first used to run cleaning

cycles, and then the HB is switched to single cycle mode while sweeping down. (If switched while sweeping up, it will hold on the forward sweep.) The instrument automatically finishes its cycle at the initial potential, so the desired hold potential is set as the initial potential. This signal is applied to the "External Signal In" connector on the Agilent Function Generator. Once HB holds at the initial potential, the signal

(55)

and record the whole period time  or the hold time 2. Then the "Output Signal" of Agilent will be applied to the "External Signal In" connector on the PAR Model 175. The "Output" mode is used to trigger PAR to reach a higher potential. By setting the high potential to be 5 V and the low potential to be 0 V, the output signal is large enough to trigger PAR at the end of the hold time. PAR generates the sweep waveform from the hold potential (hold) to the initial potential in. In more complicated cases,

PAR can generate more complicated waveforms, such as sweeping up from hold to

up, then sweeping down to inby simply changing the initial direction from down to

up. The initial potential, final potential, upper limit and lower limit potential need to be set by adding the initial potential (hold potential) from HB. For example, the initial potential in HB is 1.1 V. Then the initial potential in PAR should be set to be 0 V (11 + 0 = 11 V), the final potential should be set to be −106 V (11 − 106 = 004 V), the upper limit potential should be set to be 03 V (11 + 03 = 14 V), and the lower limit potential should be set to be −106 V (11 − 106 = 004 V). Meanwhile, the "Output Signal" of the HB and PAR are summed into a continuous sweep-hold-sweep waveform (004 V−14 V−hold− 004 V) by a Summing Amplifier in the SRS

SIM900 Mainframe. The summed signal is applied to the "External Signal In" on the Gamry potentiostat, which is operated in analog mode by running a named script. The cell is selected to be remaining on. The potentiostat applies the potential signal to the electrochemical cell, and measures the cell current and potential. The current output (out) and potential output (out) by the Gamry are reversed (−100) by the

Scaling Amplifiers in the SRS SIM900 Mainframe. The reversed potential and current are recorded by the Picotech Oscilloscope.

The process uses the following procedures: preheat instruments for half an hour, run the script in Gamry, push "INITIAL" button in PAR, then start HB to run clean cycles. While HB sweeping down, switch to single cycle mode. The  −  signal is recorded at the same time by the oscilloscope. HB generates the sweep-hold

Platinum oxide reduction kinetics on polycrystalline platinum electrodes

Geer Qile

Master of Science

Platinum Oxide Reduction Kinetics on Polycrystalline

Platinum Electrodes

Geer Qile

Abstract

Table of Contents

List of Tables

List of Figures

Nomenclature

Acknowledgements

Introduction

Chapter 2

Theories and literature review

2.1

Introduction

2.2

The electrical double layer

+

+

+

+





-working

electrode

electrolyte

(re-working

electrode

electrolyte





+

+

+

-+

j

j

j

2.3

Langmuir kinetics

(a)

(b)

2.4

Cyclic Voltammetry

2.4.1

Cyclic Voltammetry on polycrystalline Pt

2.4.2

Reaction reversibility

2.5.1

Intermediates

2.5.2

Place exchange

2.5.3

Surface structures

2.5.4

Modelling

2.6

Double-layer charge correction

Chapter 3

Techniques

3.1

Introduction

3.2

Electrochemical cell

3.3

Potentiostats

3.3.1

Digital vs Analog

3.3.2

Gamry

3.3.3

Analog potentiostat

+

I

A

B

Agilent

Gamry