PROPERTIES of PLATINIZED PLATINUM a Thesis Submitted For

Home , Chloroplatinic acid

PROPERTIES OF PLATINIZED PLATINUM

A thesis submitted for the degree of

Doctor of Philosophy

of the

University of London

Alan Martin Feltham

Department of Chemistry Imperial College of Science and Technology London S.W.7.

October, 1971 2

ABSTRACT

This thesis deals with three distinct aspects of the properties of platinized platinum.

In the preparation of platinized platinum electrodes a small amount of lead acetate is commonly added to the plating solution. The lead is partly incorporated into the platinum deposit. Its subsequent rate of leaching out has been studied, and was found to require the presence of both acid and oxygen. The results indicate that only the lead in the top 2 or 3 atomic layers of the deposit dissolves and that it is largely present there as PbO; most of the dissolution occurs within an hour. The potentials of lead-containing platinized platinum electrodes behave in a non-Nernstian manner in the presence of lead nitrate.

The major aim of platinizing platinum is to obtain a large surface area. To this end optimum plating conditions must be sought. Yet there is extensive disagreement in the literature, which is reviewed, on the way in which the area of platinized platinum electrodes varies with current density or potential of deposition. The variation of the mass specific area of platinized platinum with deposition potential from -56 to +594- mV(NHE) was tnerefore studied at constant mass degrees -2 of platinization (10 and 5 mg cm ). Maximum mass specific area was obtained at a deposition potential of approximately 150 mV(NHE) where the average current density of platinization was about 27 mA cm-2 At deposition potentials more cathodic than about 140 mV(NHE) a coupled reduction reaction led to the simultaneous evolution of hydrogen. The coulombic efficiency of the deposition process passed through a maximum at about 120 mV(NHE).

Platinum catalyses the disappearance of bromopentamminecobalt(III) bromide in acid solution. The kinetics of the corresponding catalysis by platinized platinum have been studied by means of a rotating disc system. The main catalysed reaction was found to be reduction to cobalt(II), and the presence of bromide ions seems to be essential. 3

ACKNOWLEDGEMENTS

I should like to thank Dr M. Spiro for his help and guidance throughout this work and his genuine interest in my welfare, and Professor R.M. Barrer, F.R.S., for providing research facilities.

My thanks are due also to the Science Research Council for the award of a Research Studentship.

I am grateful to Roger D. Jee for the use of his a.c. polarograph and for helpful discussions on analytical chemistry. CONTENTS Page

PART I INTRODUCTION 7

Chapter 1 Preamble. 8

1.1 Preparation of Finely Divided Platinum. 8

1.2 Historical Origins of Platinized Platinum. 8

Chapter 2 Electrode Kinetics and Mechanism of Platinization. 10 2.1 In the Absence of Lead. 12 2.2 In the Presence of Lead. 18 2.3 Substrate Pretreatment. 19

Chapter 3 Electrodeposition. 21 3.1 Deposit Appearance. 21 3.1.1 In the Absence of Lead. 21

3.1.2 In the Presence of Lead and Other Additives 22 3.2 Deposit Growth. 26

3.2.1 In the Absence of Lead. 27 3.2.2 In the Presence of Lead. 27 3.2.3 Occlusion of Reagents. 28 3.3 Deposit Structure. 28 3.4 The Effect of Alternating Current. 29

PART II THE STABILITY OF LEAD IN PLATINIZED PLATINUM DEPOSITS 31

Chapter 4 Composition of Lead Containing Deposits. 32

Chapter 5 The Leaching of Lead from Platinized Platinum Electrodes in Acid Solution. 35

5.1 Polarography. 36

5.2 Experimental Procedure. 38

5.3 Results. 50 5.4 Discussion. 54 5

Page

PART III THE SURFACE AREA OF PLATINIZED PLATINUM ELECTRODES .57

Chapter 6 The Measurement of Electrode Surface Area. 58 6.1 Definition of Useful Parameters. 58 6.2 Electrochemical Determination. 59 6.2.1 The Three Potential Regions of Platinum. 60 6.2.2 Empirical Details of Area Determination. 64

Chapter 7 Co-ordination of Past Area Work. 71

7.1 Dependence on Plating Conditions. 71

7.1.1 Degree of Platinization. 71

7.1.2 Lead Acetate Concentration. 80

7.1.3 Current Density of Deposition. 85 7.1.4 Potential of Deposition. 89 7.1.5 Temperature. 94 7.2 Reproducibility \ 94 7.3 Decrease with Time. 95

Chapter 8 The Variation of Area and Coulombic Efficiency of Platinized Platinum Electrodes with Deposition Potential at Constant Mass Degree of Platinization. 100 8.1 Experimental Procedure. 100 8.2 Results. 109 8.3 Discussion. 111 8.3.1 Mass Specific Area. 111 8.3.2 Coulombic Efficiency and Current Density. 115 8.3.3 Deposition from Different Plating Solutions. 117 8.3.4 Coulombic Specific Area. 119 8.3.5 Variation of the Parameters with Current Density of Platinization. 119 8.3.6 Hydrogen Evolution. 123 Chapter 9 Survey of Platinizing Procedures. 126

PART IV HETEROGENEOUS CATALYSIS IN SOLUTION 129

Chapter 10 Survey of Past Work. 130 10.1 Electron Transfer Reactions. 130 10.2 Isomerization Reactions. 131 10.3 Substitution Reactions. 131 6

Page Chapter 11 Theory of the Rotating Disc and Its Application to Catalysis. 134 11.1 Mass Transport in the Rotating Disc System. 135

11.1.1 Hydrodynamics. 135 11.1.2 Solute Transport. 138

11.1.3 Further Analysis. 140 11.1.4 Practical Factors. 141

11.2 Kinetic Analysis. 143

11.2.1 Surface Control. 143 11.2.2 Intermediate Kinetics. 146 Chapter 12 The Heterogeneous Catalysis of the Aquation of Bromopentamminecobalt(III) Bromide by Platinized Platinum. 150 12.1 Aims of the Present Work. 151 12.2 Preliminary Experiments. 152 12.2.1 Preparation and Analysis of Bromopentamminecobalt(III) Bromide. 152 12.2.2 The Homogeneous Rate. 153 12.2.3 Kinetic Runs with a Foil Catalyst. 154 12.3 Rotating Disc Experiments. 157. 12.3.1 Platinization. 160 12.3.2 Area Determination. 161 12.3.3 Catalysis Runs. 165

12.3.4 Production of Cobalt(II) 170

12.4 Discussion. 172 12.5 Conclusion. 175

REFERENCES 176 PART I

INTRODUCTION 8

CHAPTER I

PREAMBLE

1.1 PREPARATION OF FINELY DIVIDED PLATINUM Platinum is a good catalyst for numerous reactions. Since the catalytic activity of a heterogeneous catalyst depends greatly on its surface area, it is clearly desirable to have the platinum in as finely divided a form as possible. This can be achieved by the reduction of a platinum compound in solution to form "platinum black", and there are three ways of performing this reduction. One is chemical, e.g. the reduction of chloroplatinic acid with formaldehyde 1 in alkaline solution. Another is electrochemical, e.g. the cathodic reduction of chloroplatinic acid on platinum.2 The last method is radiolytic, e.g. the irradiation of an alkaline saturated solution of . sodium hexahydroxoplatinate(IV) with electrons .3 This thesis is concerned with the physical and chemical properties of platinum that has been electrodeposited on a platinum substrate, known as platinized platinum.

1.2 HISTORICAL ORIGINS OF PLATINIZED PLATINUM It is an accident of history that the original recipes for platinum black electrodeposition were devised, not for electrochemical purposes, but in connection with the measurement of radiation. When radiation falls on a black metallic strip its temperature rises, and the resulting increase in resistance can be determined with a Wheatstone bridge arrangement. Early models of such devices, known as bolometersi 4 made use of platinum strips blackened with soot by means of a petroleum flame. The irreproducibility of this method of blackening led Lummer 2 and Kurlbaum to try instead the electrodeposition of platinum black, since composition of the plating solution, current, voltage, and time can be exactly defined and controlled. But there was one drawback: platinum black electrodeposited from pure chloroplatinic acid solutions did not adhere properly to the electrode. In attempting to overcome 9

this difficulty, Kurlbaum and Lummer5 recalled that platinum black can easily be precipitated from chloroplatinic acid solution by adding copper or lead, so that such chemically prepared platinum black always contains some copper or lead as well. Accordingly, they tried adding a small amount of copper sulphate, to the extent of about r% of the chloroplatinic acid present, to the plating solution, and found that tnis regularly produced very good platinum black deposits. Even better and more rapid results were attained by adding a small quantity of lead acetate. Their final recipe5 was 1 part of chloroplatinic acid and 0.008 parts of lead acetate to 30 parts of water, the electrolysis being -2 carried out at a current density of 30 mA cm at the cathode and with a potential difference of 4 volts between the (platinum) working electrodes. It is this recipe which was shortly afterwards adopted by Kohlrauschb for makingplatinum black electrodes for electrochemical purposes. Kohlrausch, to whom this preparation has at times been incorrectly attributed,7 stated clearly, but without quoting any literature sources, that the method was that of Lummer and Kurlbaum. Only 8 in a later book is reference (51 specifically referred to. That this historic recipe is quite sound is shown by the results of more modern research which is discussed in the following sections.

In the original bolometric study the platinum compound was referred 2,5 10,11 to as Tlatinchlorid", but both Kohlrausch9 and more recent books have pointed out that this term was commonly applied to chloroplatinic acid. This loose nomenclature has caused some confusion in the later literature and in many present day undergraduate laboratory manuals. 12,13 Even PtCl4 may mean chloroplatinic acid unless a special preparation9 of platinic chloride is given. 10

CHAPTER 2

ELECTRODE KINETICS AND MECHANISM OF PLATINIZATION

The electrodeposition of platinum from chloroplatinic acid solution involves three couples:

IVC1 2- II Pt 6 + 2e- = Pt 2C1

PtII C1 2-4 + 2e- 4-2! Pt + "Cl

IV 2- Pt C16 4e- Pt

The electrode potentials have been measured under a variety of conditions, and the results are summarized in Table 1. Those of 14 Goldberg and Hepler are the selected best standard values. In general, the e.m.f.s of these couples lack reproducibility.15a This is not unexpecte4 reactions (1) to (3) involve the breaking of Pt-Cl bonds and must tterefore be kinetically slow and possess low exchange current densities. 2 Reference is occasionally made to the extent to which PtC16 hydrolyses in solution. The hydrolysis kinetics have been investigated 16,17,18,19,20 and the slow reaction has been found to be catalysed by 17 17 a product of the hydrolysis, light16b,17,18,19,20and platinum black. In fact, most plating solutions are acid (usually HC1 is added) and the evidence suggests strongly that under these conditions the major species present is PtC16. Only in neutral aqueous solutions of K2PtC16 or PtC14 should therm be significant concentrations of hydrolysed species. The approximate rydrolysis constants for K2PtC16 in the temperature 20 range 25 - 55°C rave been given as ICS = 6 x 10 3M and K2 = 2 x 10-4M. This means that Tor the recommended chloroplatinate concentration of 0.0718M with no added HC1, the solution will be 0.05381 in PtC16 , 0.0178M in PtC15D120.)-, and 0.0002M in FtC14(H20)2. For a solution 11

2M in HC1, the equilibrium concentrations will be 0.0716M in PtC16 1. 0.0002M in PtC15(H20)-, with a negligible amount of PtC14(H20)2.

Table 1

Electrode Potentials of the Three Pertinent Platinum Couples (see text) Eo/v a Couple T/0C Supporting Ref.. electrolyte

1 25 None 0.77 + 0.05 14 1 25 None 0.68 15b 1 25 1M HC1 0.72 b 27 1 25 1M NaC104 0.72 d 27 1 50 2.5M HC1 0.74 23 1 60 None 0.745 e 22 2 25 None 0.75 14 2 25 None 0.73 15b 2 25 1M NaCl04 0.76 b 27 2 60 None 0.785 e 22 3 25 None 0.76 e 14 3 25 None 0.705 e 15b 3 25 lM NaC1O% 0.74 b 27 3 60 None 0.765 ° 22. a All potentials are given on the NHE scale, unless stated otherwise. Values given on the SCE scale have been converted by using E(SCE) 0.245 V at 25°C. b Potentials given versus a 1M NaC1 calomel electrode (NaCE). The cell SCE/1M HC104/NaCE had an e.m.f. of 65 mV, hence E(NaCE) = 0.310 V if the large liquid junction potential is ignored. Measured versus SCE at 18°C with a KC1 salt bridge. d Calculated from the corresponding values for couples (2) and (3). e Calculated from the corresponding values for couples (1) and (2), 12

2.1 IN THE ABSENCE OF LEAD The earliest electrochemical studies of the plating of platinum 2 22,25 from PtC16 solutions appeared more than a third of a century , 2,516,8 2 after the process itself had been introducea. That PtC1 participated was shown by two main lines of evidence. Firs::, the current-voltage curves, of which an example is given in Figure 1, exhibited not one but several waves, the last steep rise being caused by hydrogen evolution. The quantitative concordance between the 22,25 24 25 curves obtained by early and later / workers is relatively poor, partly because the compositions and temperatures of the plating solutions varied and also as a result of the different speeds with which the curves were drawn. The slower the experiment, the greater the area of the platinized platinum deposit and so the lower the real 22 current density. Grube and Reinhardt's results are not untypical and encompass a wider range of conditions than most: their initial deposition potentials with 0,1M H2PtC16 at 18°C were 0.45 V in the absence of HC1 and 0.12 V in 5M HC1 and, at 60°C, 0.54 V and 0.36 V, respectively. Comparison with Table 1 shows that the initial electrode process (almost certainly reaction (1)) is irreversible and this, as has already been pointed out, is not unexpected. The shift to more anodic deposition potentials as the temperature is raised demonstrates the anticipated increase in exchange current density. 2- The second reason for pointing to the participation of PtC14 came from the analysis of the electrode products after various times. 2 Initially there was preferential reduction to PtC14 , as shown also by the fact that the solution colour turned from lemon yellow to dark red, to be followed by the deposition of platinum metal. This change 22 with time was particularly pronounced at higher temperatures.

Another third of a century passed before more quantitative information appeared with the advent of sophisticated electrochemical techniques. The most useful of these to date has been thin-layer voltammetry 26 where the peak currents are a few microamperes only and the amount of 21 platinum deposited during the course of each run is manageably small. 27 Lau and Hubbard applied slow voltage sweeps to a cell 26 pm thick 2- 2- containing either PtC16 or PtC14 solutions, and fitted the resulting Figure 1. Current (I)'- voltage (E) curves for the platinization of platinum from a .0.05M chloroplptinic acid solution with no added HC1; no lead acetate and, 2.7 x 10-,M lead acetate ----. The area of the electrode was not given; from Ref.25.

1.0 .0104,mommapli

../ 0.0

0.8 0.6 0.4 02 E ly 14

current-voltage'curves to theoretical rate equations. The derived rate constants k are listed in Table 2. In agreement with most e.m.f. and the early voltammetry evidence, both reactions (1) amd (2) are seen to be electrochemically irreversible. The addition of chloride has an adverse effect, to such an extent that the reduction 2 of PtC14 to platinum in 1M chloride solutions takes place at potentials so cathodic that solvent reduction occurs simultaneously. At low 2 chloride concentrations, on the other hand, conversion of PtC16 to 2 a PtC14 is rapidly followed by the latterls reduction to the metal, and the electrode process appears to be simply reaction (3). The reaction

2 2- PtC16 Pt 2C1 = 2 PtC14 ( 4)

should play a negligible role since it took 500 hours at 60C to come 22 to equilibrium.

The following discussion is based on the values of the rate constants as listed in Table 2, although their magnitudes and particularly -the trends they display depend strongly on the E° assignments (see footnote a). As the chloride concentration rises, ki decreases at first and then reaches a steady value whereas k2 decreases more and more rapidly. Their behaviour is therefore qualitatively different. 2 2- The potentials of reduction of PtC16 and PtC14 in low chloride media are similar and the charges of the ions are identical: their shapes, 2- 2 however, are not, PtC16 being octahedral and PtC14 planar. This fact has already been invoked to explain why, at mercury cathodes, the 2 reduction of PtC16 (and of many other large anions) is much more - sensitive to the presence of cations than is that of PtC14 .

Table 2 a 2- 2- Rate Parameters for the reduction of PtC16 to PtC1, - (ki , Mi ni ) and of PtCl2 to Platinum (k, , a, n,°) at 25°C from Ref.27.

NaC1 concn, lc, /cm s-1 Q2 n,23 k2 /cm s-1 M b rli0 -6 -6 0 0.29 6.1 x 10 0.38 8.1 x 10 0.01 0.37 3.5 x 10-6 0.32 3.4 x 10-6 -6 0.10 0.37 1.5 x 10 0.32 1.6 x 10 7 0.30 0.45 5.9 x 10-7 - 's 2 x 10-9 7 1.00 0.44 5.7 x 10 - < 1 x lo 9

The parameters were derived from the basic equation o i = nFkC exp [- an F(E - E°)/RT] ox relating current density i to potential E. The E° terms were taken as standard and not as formal potentials, the values used being = 0.40 (NaCE) and = 0.45 (NaCE) V at all chloride concentrations.

In addition to the NaC1 concentration indicated, the solutions 2- 2- contained 1 mM PtC16 or PtC14 , 0.01M IIC104 , and sufficient NaC104 to adjust the ionic strength to 1.01. 28 If, as Frumkin has suggested, the transition states of the flat 2- - PtC14 ions are located close to the electrode surface while the PtC16 ions to be discharged are linked to it by cationic bridges, the behaviour at mercury electrodes becomes comprehensible. A similar situation may 2- obtain at platinum. A flat PtC14 ion will then require a group of at least five adjoining surface atoms to be free of adsorbed chloride for discharge to be possible, and this becomes increasingly difficult 27 as the chloride concentration rises. Hence the sharp fall in the values of k2. Another test of the Frumkin hypothesis could be made by measuring k1 and k2 in the presence of supporting electrolytes containing different cations. Some information on this aspect is already available. According to Table 2, when Na+ is almost the sole counter ion, k, is constant as the chloride ion concentration changes 29 from 0.3 to 1M whereas, in the presence of 3M H2SO4 , k1 varies inversely with sodium chloride concentration over the range.0.4 - 1M. In the 29 original paper the authors deduced from .the latter finding that the step determining the rate of the cathodic process involves the complex PtC15 , formed by the preceding reversible reaction

2- PtC16 PtC15 + Cl (5)

27 However, the rate of a reaction such as (5) or the equivalent reaction producing Pt(H20)C15 is too small to account for the observed rates of reduction.

Table 2 records also the values of Cori the product of the cathodic transfer coefficient and the number of faradays involved in the formation 27 of 1 mol of activated complex. Although it did not prove possible o to separate these two parameters, the results suggest that both ni and o n2 are unity. This implies that transient Pt(III) and Pt(I) species, respectively, are produced in the electrode reaction. Lingane,30 in 2 an earlier chronopotentiometric study of PtC16 reduction kinetics in IM HC1 on slightly platinized electrodes, found the curve to fit, very roughly, a diffusion-controlled one-electron process. He interpreted this as an initial reduction of Pt(IV) to Pt(III), followed by the latter's disproportionation to Pt(IV) and Pt(II). Although the evidence is as yet far from convincing, it is worth noting that short-lived 17

31,32 Pt(III) and Pt(I) species have recently been formed in aqueous solutions by electron pulse radiolysis.

There is considerable evidence in the literature that the platinizing process is a more complex one than the thin-layer voltammetry results in Table 2 lead one to suppose. In Lingane's chronopotentio- 2 metric experiments,3° no discrete reduction wave for PtC16 in 1M HC1 occurred on ucleann platinum electrodes but only on those covered with a visible coating of platinum black. Lingane concluded that metallic platinum must participate in the reduction process in a much more specific way than merely serving as an inert electron source, and proposed an initial cnemical reduction of the oxidant by platinum metal. Another chronopotentiometric finding vias an ageing effect. If the slightly platinized electrode was allowed to stand in the de-oxygenated 1M HC1 test solution for two days (a medium in which the extent of hydrolysis should be slight), a four-electron reduction wave was recorded, corresponding to reaction (3). This is in marked contrast to the thin- layer results which showed the reduction steps Pt(IV) Pt(II) and Pt(II) -4 Pt to be well separated in 1M chloride solutions. Whether the pH difference of 2 between the solutions of Lingane (pit 0) and those of Lau and Hubbard (pH Pe2) is in part responsible for these considerable differences is not known, but it does seem clear that the state of the surface plays a major role in the electrode kinetics.

Two other puzzling observations in the early literature bear this out. Certain curious behaviour, such as strong potential fluctuations 22 in one region of the current-voltage curve, was said to point to the formation at the electrode surface of a poorly conducting layer which changed as the electrolysis continued. A peak at the beginning of a 23 cnronopotentiogram was also interpreted in this way although it is more likely to have been caused by slow nucleation.33 A second unexpected phenomenon during platinization was the evolution of hydrogen at potentials more positive by up to 0.14 V than that calculated for 22,23 30 the reversible hydrogen electrode. Lingane, too, obserVed this 2- A-, and attributed it to simultaneous reduction of PtC16 and H the latter forming hydrogen at a very low partial pressure, This phenomenon is discussed in more detail in Part III. It is interesting to recall that long ago Kohlrausch9 suggested that at high current densities, 3.8

solutions of PtCl4 (not H2PtC16 ) produced platinum at the cathode by the formation of hydrogen which subsequently reduced the platinic chloride. Further study in this area should prove interesting.

2.2 IN THE PRESENCE OF LEAD In the preceding section we have seen how little is understood of the electrode kinetics of platinum deposition from chloroplatinic acid solutions. Considerably less is known of platinum deposition from lead-containing solutions, an astonishing fact when we bear in mind that the process is now three-quarters of a century old and is in daily use around the world. 25 Virtually the first attempt at an electrochemical study was in 1970, and it produced some unexpected results. First, a solution containing no platinum but 0601M in lead acetate and 0.1M in HC1 showed a reduction wave on a platinized platinum electrode beginning at +0.74 V although the standard potential of the Pb"/Pb electrode15c is -0.126 V. This is almost certainly related to the fact that a lead-containing platinized platinum electrode in a lead ion solution exhibits a potential of ca. 0.83 V (see Chapter 5). This potential does not obey the Nernst equation and the facts suggest a very low lead activity coefficient in the deposit indicative either of strong lead-platinum bonding or steric or diffusion-limited imprisonment of lead in the platinum lattice.

The presence of small amounts (10 4 - 10-3M) of lead acetate in 25 chloroplatinic acid plating solutions strongly affects the cathodic current-voltage curve (cf Figure 1). The curve alters shape, shifts anodically by 100-200 mV, and the current at a given potential increases. Lead therefore decreases the overpotential, and increases the rate, 2 34 of the reduction of PtC16 . This, incidentally, contradicts a theory that states that, in cases of co-deposition, the rate of reduction is lowered for the metal with the more positive value of the potential of zero charge. E is + 0.15 V for platinum and - 0.65 V for lead.35 pze 25 Bernard believes, on rather skimpy evidence, that the lead is not deposited concurrently on tne platinum but is adsorbed on the electrode. In this way it inhibits the growth of crystals and favours the creation of new crystallites, but it is not clear now the lead comes to be incorporated into the deposit. Electrocrystallization 19

studies coupled with structural probing would prove most enlightening. 25 The other function of the lead could be inhibition of the evolution of hydrogen and so an increase in the coulombic efficiency of platinum deposition.

2.3 SUBSTRATE PRETREATMENT

Platinized platinum deposits formed on untreated platinum substrate electrodes were more fragile3b than those formed on annealed or sand- 36 blasted surfaces. Electron microscopy showed that a smooth platinum surface was roughened, to an increasing degree, by thermal etching, sandblasting, and etching with aqua regia. On a roughened surface the initial real current density will be lower than for a smooth one, at constant geometric current density. This is likely to produce more adherent deposits, as is well known in electrofinishing. The other effect of these treatments, particularly etching by heat or with aqua regia, is to remove impurities% such as grease, from the substrate surface.

Some interesting information is provided by a nucleation study37 of mercury on platinum single crystals, Here, platinum was anodized to intense oxygen evolution in mercurous nitrate solution and subsequently soaked in 0.09 - 1.311 nitric acid for 5 - 60 s. The electrolytic nucleation of mercury from mercurous nitrate solution was then observed. The critical overpotential for nucleus formation diminished substantially with increasing time of soaking and nitric acid concentration. Cathodic prepolarization in nitric acid after the anodization produced an even more active electrode, and it was concluded that the dilute nitric acid removed the surface oxide layer. At concentrations below about 2M, 38 nitric acid has virtually no oxidizing power and there is the possi- 59a bility that the Pt-0 layer is unstable in acid solution in the absence of dissolved oxygen. However, in the nucleation work37 no mention was made of de-aeration.

The pre-treatment that proved most effective for deposition of mercury on platinum, namely anodization followed by cathodization, is 40 similar to that conventionally recommended for the platinization of platinum. This is, in brief, washing the platinum substrate in either warm concentrated nitric acid or in aqua regia followed by nitric acid, and subsequent cathodization in very dilute sulphuric 20

acid. There is soma foundation for this procedure. Concentrated nitric acidproduces an oxide film on the platinum surface,39a and 3b9 some workers believe that cathodization of a pre-anodized electrode (i.e. oxide-covered) produces an invisible film of platinum black. Lingane30 has shown (Chapter 2.1) that the formation of such a film - is essential for rapid reduction of PtC16 . The electrode kinetics 2 of PtC16 reduction on platinum substrates subjected to various electrochemical nre-treatments have not yet been studied. It is known that the exchange current densities of a few couples are greater on pre-anodized platinum surfaces while those of most couples are higher on pre-cathodized surfaces.39b,41,42 21

CHAPTER 3

ELECTRODEPOSITION

No thorough electrocrystalIisation study has been carried out. In general, the work that has been done has been purely empirical in approach. Rather more attention has been paid to platinization from solutions containing a metal additive, usually lead acetate.

3.1 DEPOSIT APPEARANCE

Nearly all papers in which platinized platinum electrodes are employed describe their appearance. The presence of additives in the plating solution can have a startling effect and the range of textures obtainable is wide.

3.1.1 IN THE ABSENCE OF LEAD Platinization with solutions containing no additive has always 2,7,12,43,44 given grey deposits. In many cases these were not adherent and tended to flake off. Adherent deposits have, however, been prepared from solutions containing no lead acetate by one of the following procedures; platinizing only slightly? (non-adherency 45,46 occurred after intense platinization); employing a low current density (10 mA cm-2); or by holding the platinum cathode47 at +50 mV, instead of the more usual galvanostatic conditions. It is interesting to note 46 that a common factor in all but one of these procedures was the addition of HC1 to the platinizing solutions. The first mentioned? employed 0.025M HC1 and obtained only moderately good deposits, the 45 47 other two used 2M HC1 and 1M HC1 and completely satisfactory deposits resulted. This might be due to the kinetic inhibitory effect of chloride. Indeed, in cnloroplatinic acid solutions 6.0 - d 48 10.7M in HCI smooth ductile platinum deposits can be obtaine .

A similar inhibitory effect occurs" with the additive cetyl ammonium bromide, and leads to highly uniform smooth deposits. Bright 52 platinum deposits can be obtained from solutions of chloroplatinous 22

acid. The possibility of obtaining bright deposits51 from solutions of chloroplatinic acid, or better, alkaline chloroplatinate solutions, was later disclaimed50 and attributed to the presence of chloroplatinous acid produced as a result of overheating the platinic compound during its preparation. This effect of cnloroplatinous acid on chloroplatinic 48 acid plating solutions was later noted again. but just why bright 2- deposits should result under these circumstances when PtC14 is an intermediate in platinum black plating is not at all clear, and should be looked into. The commercial systems for bright platinum plating 52 on to noble and base metal substrates have been reviewed.

3.1.2 IN THE PRESENCE OF LEAD AND OTHER ADDITIVES Adherent deposits, usually black, are best produced by adding lead acetate to the platinizing solution, as Kurlbaum and Lummer discovered5 long ago. Other additives have been studied but,none has been employed as extensively as lead acetate. Copper5,53 and mercury53 were found to be acceptable substitutes for lead, and gold and thallium gave12 deposits of good quality. Cadmium, zinc, nickel, and iron, have 12,53 given grey inferior deposits. The effects of a host of metal additives on the mppearance and on the X-ray diffraction patterns of 54 platinized platinme deposits have been examined (see later).

The final appearance and texture of a platinized platinum electrode depends on the concentration of lead acetate in the plating solution and the current density of deposition. The minimum concentration of lead acetate required to give a good black deposit from a 0.062M (3%) -2 43 chloroplatinic acid solution 0.5M in HC1 at 13.5 mA cm was found to be 2.5 x 10-4M (0.01%). A black, velvety texture is usually associated with lead-containing platinized platinum electrodes, although at higher lead acetate concentrations they become grey and smooth (Table 3). Kurlbaum,55 speculating as to why platinized platinum deposits were black, compared them to an end-on view of a collection of aligned shiny knitting needles. Internal reflection reduces net reflection. Experimental support for this picturesque illustration is apparent in Table 3, where the blackest deposits have the highest roughness factors, i.e. the highest areas. 23

Table 3

Data on Platinization for varying Lead Acetate Concentrations a

Conc. of, lead Appearance Coulombic Pb/Pt Roughness acetate of deposit efficiency atomic factor % w/v 104m ratio

0 0 Grey, compact 0 358 0.003 0.8 Black, very powdery 40.6 0.0021 534 0.01 2.6 Black, very powdery 35.9 0.0041 509 0.05 13 Black, powdery 37.2 0.0143 261 0.075 20 Grey, compact 41.6 0.0208 229 0.1 26 Grey, compact 41.6 0.0230 296 0.15 40 Grey, compact 39.8 0.0247 0.2 53 Grey, compact 38.4 0.0256 296

a Using 0.051M (2.5%) chloroplatinic acid at 10 mA cm-2 for 1 hour with current reversal every 1.5 minutes; from Ref.46. 24

Table 4

Variation of appearance of Platinized Platinum Electrodes with Lead Acetate concentration a

Conc. of lead acetate Appearance of deposit 4 % w/v 10 M

0 0 Dark grey, compact 0.002 0.5 Black, compact 0.01 2.6 Black, compact ` 0.02 5.3 Black, compact 0.05 13 Black, very powdery 0.08 21 Black, very powdery 0.1 26 Black, very powdery 0.15 40 Black, very powdery

a Using 0.021M (1%) chloroplatinic acid in 0.1M HC1 at 625 mA cm-2 for 10 minutes; from Ref.54.

b The deposits were black and compact up to 625 mA cm-2 but black and powdery with current densities in the range -2 850 - 1250 mA cm . 25

Comparison of Tables 3 and 4 suggests that the appearance of the platinized platinum deposit also depends on the current density of platinization. Footnote b in Table 4 confirms this. The powdery deposits obtained at high current densities, and therefore at rather cathodic potentials, are no doubt attributable in part to pitting by the simultaneous hydrogen evolution (this could perhaps be reduced by strong stirring). However, this conclusion seems oddly at variance with the findings in Table 3 where a very low..current density was employed and yet, at a lead acetate concentration of 5.3 x 10-4M (0.02%) (the same as in footnote b of Table 4), a black and powdery deposit was found. A possible explanation is that the current reversal used in Table 3 was harmful and the chlorine evolved caused , gas pitting, although it has been founa43 -,45 that current reversal 56 has no effect. Potter, on the other hand, claims it is beneficial and decreases the amount of occluded gases. An alternative suggestion is that the formation of a black compact deposit at 5.3 x 10-4m (0.02%) lead acetate in Table 4 but not Table 3 is a consequence of the nigher lead acetate/chloroplatinic acid ratio in the plating solution. Yet this idea, too, is inadequate, for Table 4 shows that black compact deposits are produced at low rather than at high lead acetate/cnloroplaZinic acid ratios, at least at a very high current density.

The information on appearance in Tables 3 and 4 is supplemented by isolated observations. Deposits obtained3b from a 0.103M (5%) chloroplatinic acid plus 2 x 10 3M (0.076%) lead acetate solution at 10 mA cm-2 were smooth and grey (consistent with Table 3) but at -2 100 mA cm they were coarser in appearance and characteristic of a dendritic structure. Similar behaviour was observed for a 0.072M (3.5%) chloroplatinic acid plus. 5.2 x 10-3m (0.2%) lead acetate solution,0.1M in HC1, on changing the current density from 10 to -2 30mA cm . Yet with 0.072M chloroplatinic acid, 2M HC1, and -2 1.3 x 10 4M (0.005%) lead acetate, 30 mA cm gave a velvety black deposit (see Chapter 4). Clearly further systematic work linking appearance with current density, solution composition, and other variables is required before any systematic relationship can be formulated. 26

Most appearance studies were carried out at constant current, but at least one report57'refers to potentiostatic work. In this, 4 using a solution 0.05M (2.4%) in chloroplatinic acid plus 2.7 x 10 m (0.01%) in lead acetate, deposits obtained at potentials more anodic than +395 mV (very low current densities) were grey, while those obtained at still more anodic potentials were not adherent or reproducible. Deposits formed at potentials slightly more cathodic than +145 mV (high current densities) were finely divided and very black. This is essentially in agreement with the work described in Chapter 8.

Appearance is the most obvious and direct way of characterizing 6 the texture of a surface, but Kohlrausch also eXamined the wettability of black platinized platinum electrodes prepared according to Kurlbaum and Lummer's recipe.5 He found that generally they were easily and well wetted with water. However, a large lightly platinized electrode became de-activated when allowed to become dry and was then no longer wetted by water. Optical experiments indicated the presence of an air layer between the platinum particles. The wettability was restored by adding a drop of alcohol to the electrode. In contrast, the smooth grey electrodes prepared from solutions nigh in lead shed water quite readily. The capillary nature of the black deposits is obviously responsible for their enhanced wettability.

3.2 DEPOSIT GROWTH

A most important aspect of platinization that has yet to be properly investigated is the nucleation and growth of the deposit. 58 Electrocrystallization studies have been carried out on the deposition of platinum on mercury (where it forms an amalgamated two-dimensional layer which changes to a three-dimensional structure as the Pt(IV) -4 concentration exceeds 10 M) and on glassy carbon. Here thick platinum deposits grew, the nuclei being essentially hemispherical but with small protruding dendrites which points to mass transport control. Growth of platinum on platinum might well follow a similar path to that on a carbon substrate,59 but little direct work has been done. 27

3.2.1 IN THE ABSENCE OF LEAD Electrocrystallisation studies in general nave snown33/6° that rates of nucleation are strongly dependent on overpotential. On a surface with a high density of dislocations 1010 cm-2), no 6o nucleation is possible unless cathodic overpotentials exceed 150 mV. At overpotentials below this value the growth occurs at the dislocations. Platinization on platinum requires high cathodic overpotentials 36,55,61 (Chapter , 2.1) and indeed nucleation has been favoured as the mode of deposit growth. Most experimental work, however, has been done galvanostatically. ,36 Deposits on platinum obtainea from a 0.041M (2%) chloroplatinic -2 acid solution at 100 mA cm were found by replication electron microscopy to form as randomly distributed particles which grew laterally to cover the electrode surface. The same general growth pattern occurred for various substrate topographies (untreated, metallograpnically polished, sandblasted, and etched by heat, molten sodium carbonate and aqua regia) but with minor variations. On an untreated platinum substrate the deposit thickness growth rate was not linear and obeyed the equation 8 d = Dt (6) where d is the thickness at time, t, and D and 8 are constants whose 36 values were not given. Optical microscopy revealed these surfaces to be rough and 7.11thout any distinguishing features, and cross-sectional microscopy snowell that they appeared as solid films with vertical cracks. The latter technique was also applied to similarly prepared deposits at -2 10 mA cm and the same features emerged. The deposit thickness growth -4 -1 rate on an untreated platinum substrate was 2 x 10 mm s .

3.2.2 IN THE PRESENCE OF LEAD 36 It has been shown by electron and optical microscopy that more nuclei were formed from solutions containing lead than those without. In the same work, cross-sectional studies revealed a transition from 4 a closely packed deposit prepared from a solution of 1 x 10 m (0.0038%) lead acetate to a dendritic type deposit from one of -2 1 x 10 M (0.38%) lead acetate (using 0.1M (4.9%) chloroplatinic 28

acid and 100 mA cm-2). The visual texture changed correspondingly from fine grains to a coarse open structure. -2 36 At 100 mA cm a 2 x 10 3m (0.076%) lead acetate solution gave -1 a constant deposit thickness growth rate of 2.8 x 10-3mm s . At -2 10 mA cm this solution produced rapid initial growth followed by a linear one after approximately 200 s. In another study, b2a 0.103M (5%) chloroplatinic acid plus 5.3 x 10-4m (0.02%) lead acetate solution at -2 -2 -1 120 mA cm gave a deposition rate of 2.8 mg cm min which is -1 equivalent to 2 x 105mm s if the density of platinum is taken as 21.45 g cm-3. This'suggests that a small change in plating conditions causes a vast change either in cculombic efficiency or, less likely, in the density of the deposit. A repetition of these growth experiments under potentiostatic conditions should clarify the situation.

3.2.3 OCCLUSION OF REAGENTS 6 Kohlrausch observed that after platinization several days of washing was required before the conductivity of the rinsing water stabilized. If a thoroughly washed platinized platinum electrode was immersed for a long time in platinizing solution the phenomenon was not repeated, so Kohlrausch suggested that the solute was occluded only during platinization. Similarly, there was an overwhelming tendency51 to occlude acid during platinization and subsequently, if in neutral solution, to emit such acids slowly but continuously. The trouble caused by solute adsorption on platinized platinum used 4o for a hydrogen electrode in unbuffered solutions has been discussed.

3.3 DEPOSIT STRUCTURE

Platinized platinum deposits containing lead have been examined with X-rays. Debye-Scherrer diffraction patterns were obtained, indicating the deposits to be crystalline, for deposits formed from 12 , 46 solutions of up to 2.9 x 10 3m (0.11%) and 5.3 x 10-3m (0.2%) lead acetate, with 0.062M (3%) and 0.051M (2.5%) chloroplatinic acid -2 54 respectively, at 10 mA cm 0 Lead acetate concentrations below -2 2.1 x 10-3m (0.08%), with 0.021M (1%) chloroplatinic acid at 625 mA cm 4 also gave deposits which exhibited Debye-Scherrer patterns, but when, at this high current density, higher lead acetate concentrations of 29

2.1 x 10-3M (0.08%) to 4.O x 10-3M (0.15%) were used, no diffraction lines were observed. This was attributed to the deposit being in a state of colloidal dispersion. 46 54 In these diffractionpatterns ' an appreciable line broadening, which tended to increase with lead content, suggested that the crystallite sizes were smaller in the higher lead-containing deposits. One would 63 suppose, therefore, that the latter possessed larger specific areas, yet the reverse is true for the higher lead contents (see Figure 16). The situation is obviously more complicated than appears at first sight, 64 and indeed, electron microscopy indicated that there is not a continuous distribution of crystallite size about a mean value, but rather two cubic types of crystallite, each with a distribution of size about its own mean value. The lattice constant of the lead-containing deposits46,54 was 46 greater than that of pure platinum, but it did not increase with an increase in the lead content.

Other metal additives that produced a widening of the platinum • crystal lattice54 in the electrodeposit were Hg, Cd and Ti, whilst Cr, Mn, Fe, Co, Ni, Cu, Zn, and Pd caused a contraction, and Sb, Sn, Di, As and Au, although they modified the nature of the deposit, did not alter the platinum lattice dimensions. No correlation can be seen between the effect of these metals on the lattice dimensions and the appearances of the deposits discussed earlier, brought about by their inclusion in the plating solution.

3.4 THE EFFECT OF ALTERNATING CURRENT

Platinum black surfaces can be produced on smooth shiny platinum in an inert aqueous solution by several minutes of a.c. polarization. 65 The conditions used have been 0.5M sulphuric acid at 60 Hz and with an amplitude sufficient to reduce and oxidise the surface without substantial gas evolution; after such polarization platinum has also 66 67 been found in solution. Others have used either 1M perchloric acid or 3M nitric acid with a 30 Hz square wave of 500 mV amplitude centered on 950 mV. Here the reflectance of the electrode began to decrease before visible darkening, and the time of blackening shortened with increasing frequency. Strangely, it was only in alkaline solution that darkening of the electrode was first noted,68 30

on the application of a square wave pulse. Moreover, under certain conditions, with the application of a.c.•a roughened electrode can be 66,69 smoothed.

There is controversy over the mechanism of this darkening. On 39b the one hand it has been attributed to the repeated penetration and removal of hydrogen caused by the cathodic parts of the pulses with the resultant expansion and contraction of the platinum lattice 66,69 breaking up the metal surface. Other workers have focussed 66 attention on the anodic parts of the pulses, and it has been suggested that the formation of strong Pt-O chemisorption bonds weakens the 70 platinum - platinum interactions. A relevant observation is that the electrolytic formation and reduction of thick bulk oxide layers, repeated several times, increased the roughness factor of a platinum electrode by 100 times or more. Perhaps both mecnanisms are admissible under appropriate conditions. 31

PART II

THE 'STABILITY OF LEAD IN PLATINIZED PLATINUM DEPOSITS 32

CHAPTER 4

COMPOSITION OF LEAD CONTAINING DEPOSITS

As the concentration of lead acetate in the plating solution increases, so does the amount of lead included in the platinized platinum deposit. The available data are plotted in Figure 2, 12 from the radiotracer work of Hevesy and Somiya and the non- 46 destructive X-ray fluorescence spectroscopy of Thacker. The latter's figures are also incorporated in Table 3. The striking disagreement in the diagram is somewnat difficult to explain. The only apparent difference in preparation was that Thacker used current reversal, whereas Hevesy and Somiya did not. Since the anodization potential (on current reversal) is probably high enough to impede loss of lead in the acid solution, the reason almost certainly does not lie here. A very puzzling feature of Hevesy and Somiya's paper is that they claimed to nave measured the lead content of deposits by the a-particle activity of a thorium B k'- 212r b, indicator, but thorium B does not emit a-particles and is a P-particle emitter. The only lead 210 a-particle emitter is radium D ( Pb). Values of lead content were estimated by comparison with lead dioxide containing the tracer, and multiplying by, amongst other quantities, the inverse of the ratio of the a-particle ranges in the two solids. Thus, change in this ratio can vastly affect the final results. It would therefore seem that Thacker's results are the more trustworthy of the two, 36 In another investigation it was found that the lead contents of the deposits obtained from a 0.1M (4.9%) chloroplatinic acid plus 2 x 10-3M (0.076)) lead acetate solution passed through a maximum as the current density i increased from 0 to 100 mA cm-2, the total amount of electricity being kept constant'by adjusting the time t. Equation (7) expresses the results:

Pb in deposit 2.= kt (i/t) (7) Figure 2. Dependence of the lead content of a platinized platinum deposit on the lead acetate concentration in the.plating solution, at 10 mA cm-2

0.06 ReF. 12.; 0.0621-1 (3 %) chloroplatirac acid in a 2 vi hydrochloric acid. Time of plairtizoLion was not given, and the current density was stated to be 10cm1) probably a misprint For 10 real cra-1.

r' ReP.4G; 0.051 M (2.570 chloroplotinic acid ) curreviL. passed For I hour with reversal every 1•6 minutes.

•awal•••••4•110

1 2 3. Lecui AcebQh CortcaliErablor‘ 31+

The magnitudes of the constants k and g were not stated, nor was the method of estimating the lead contents of the deposits.

Indications as to the physicochemical state of the lead in platinized platinum deposits come from several directions. From X-ray diffraction work it has been deduced both that the lead is46'54 12 in solid solution, and that it is not in solid solution. The equilibrium platinum-lead phase diagram71 suggests that all the lead is present as Pt3Pb dispersed in a matrix of Pt; while the e.m.f. versus composition curves72 of cast alloys indicate a compound of stoichiometry PbPt in Pt, and this has been verified72 by microscopy. However, it is essential to point out that34 the phase state of alloys produced by electrolysis often does not correspond to that of the equilibrium phase diagram. 12 Hevesy and Somiya examined the retention of lead by electrodes at high temperatures using both the a(?)-particle and y-ray activity of the thorium B tracer. According to the a-particle activity the lead started to sublime out of the deposit at about 600°C and 85% of it had been removed after 16 hours at 720°C. However, the y-ray activity indicated that only 42% of the lead had been lost after 16 hours at 690°c. Since y-rays are more penetrating than a-particles it was concluded that lead had vaporised only from a surface region, as deep as the range of the a-particles in the deposit. This would be expected, as the rate of sublimation of lead from the bulk of the deposit would be limited by the slow rate of diffusion. A similar situation pertains to the leaching of lead from platinized platinum electrodes by 1M perchloric acid, as will shortly be discussed. 35

CHAPTER 5

THE LEACHING OF LEAD FROM PLATINIZED PLATINUM ELECTRODES IN ACID SOLUTION

Lead is not thermodynamically stable in acid solution as is shown by the large negative standard Gibbs free energy changes73 of the two obvious reactions: + -1 Pb + 2H -, Pb+ H2 be = -24.3 kJ mol (8) 2+ Pb + 2H++ /20 2 -, Pb + H2O ; LIG° = -261.5 kJ mo1-1 (9) o However, these values are not strictly applicable to the dissolution of lead from the platinum electrodeposit where its activity cannot be 72 taken as unity. Indeed, the report that the e.m.f. of the cell • Pb 10.5M Pb(NO3)2 Pb-Rt (> 50 atom % Pt) was 720-740 mV suggests that the activity of lead is extremely small, although no evidence was given as to the electrochemical reversibility of the cast alloy electrode. Even more important is the absence of information in the literature as to the rate of dissolution of lead from a platinized platinum electrode. This lack of knowledge as to the thermodynamic and kinetic stability of lead in platinized platinum deposits, coupled with the latter's widespread use, prompted the present study.

Two types of deposit, corresponding to extremes of lead content, were examined, A low-lead electrode was prepared using optimal conditions for obtaining a large area (Part III), and a high-lead deposit was made with the highest lead acetate concentration previously used." E.m.f. measurements and kinetic leaching experiments were carried out with both kinds of electrode.

Perchloric acid was used to leach the lead from the electrodes, and very low lead concentrations were to be expected. For the trace analysis of lead there are basically two techniques; namely, atomic absorption and polarography. By standard atomic absorption techniques the sensitivity for lead determination74 is 0.7 p.p.m. (3.4 x 10-6M) at 36

285...,nm. Classical d.c. polarography generally has a lower limit 75 -5 of sensitivity of 10 M for metal ions, while a.c. polarography can 75 -6 give 10 or even 10 7M. In the particular case of lead, measurements as low as 0.1 p.p.m. (4.8 x 10 7M) and even 0.026 p.p.m. (1.3 x 10 7M) 76 have been claimed for a.c. polarography. For reasons of better sensitivity and the fact that the equipment was more readily available, a.c. polarography was selected for the analyses. This technique will now be briefly described as well as its relationship to classical d.c. polarography.

5.1 POLAROGRAPHY

Polarography consists75 essentially of determining (usually cathodic) current-voltage curves with a dropping mercury electrode (d.m.e.), and identifying the waves as corresponding to specific ions. In classical d.c. polarography this is done by applying a continually and slowly varying voltage between the d.m.e. and a mercury pool in the same solution, or some other counter electrode, and measuring the current between them. For the analysis of metal ions the d.m.e. is made cathodic and to suppress electrical migration of the ions of concern, a supporting inert electrolyte is added75 to the solution in at least 50 - 100 times excess. Oxygen must be removed from the solution (by purging with nitrogen) since this would also be reduced. The current varies during tne life-time of the drop, and the mean current is usually measured. For a simple fast process wnere the current is limited by diffusion and is small enough to have a negligible effect on concentration, the mean limiting diffusion current (Id/pA) is given by75 the Ilkovic equation:

I = 607 nEl2m3 c (10) d

where n is the number of electrons involved in the reduction, D the 2 diffusion coefficient of the ion in the solution (cm s-1), m the rate -1 of flow of mercury (mg s ) the drop time (s) which is the time taken for a mercury drop to grow to its maximum size (a constant), and c the .-1% concentration of the electroactive species (mol ). To this current must be added75 a small contribution for charging the double layer at the d.m.e.-solution interface - the residual current. The wave of

a particular species is characterised by the half-wave potential (Et), -2 which is the potential at the mid-point of the wave where the current is half the limiting value. For the reduction of a metal ion to form an amalgam, which is of interest here, the half-wave potential is given76 by:

El = E°' (RT/nF) In (DI/D) 2, 2 where E°'1 is the formal potential of the redox couple, and D' the diffusion coefficient of the metal in the amalgam.

'With a.c. polarography75,76 an alternating voltage is superimposed onto the polarising potential, and the alternating current is measured. The amplitude of the alternating voltage signal applied must be much less than the voltage range covered in the whole sweep. At values of the direct potential where a redox reaction occurs, the effect of a superimposed sinusoidal alternating voltage is to produce periodic concentration changes of oxidant and reductant at the electrode-solution interface. These concentration changes are accompanied by periodic diffusion processes and the flow of alternating current - the faradaic alternating current. 76 The mathematical treatment of the a.c. currents is only valid for reversible systems and, indeed, there is evidence that a.c. polarographic waves are generally only observed when the d.c. polarographic process is reversible. Thus oxygen, whose reduction at mercury is irreversible, need not be excluded. At low frequencies the periodic rate of the electrochemical reaction at the interface is very much 76 faster than the periodic diffusion changes. This means that the important parameters are diffusion coefficients, as opposed to electron transfer constants, and the use of low frequencies is best for analytical use. The instantaneous amplitude of the alternating current (Al) at 76 low frequencies is given by :

-n2F2AVe ( WD') 2 exp 1 nF(E-E°.1)7RT) dI - (12) RT[(D1/D)"-1- exp {nF(E-E°1)/RT}]2

" where A is the time-dependent area of the mercury drop (cc m3 t3 ), V the amplitude of the alternating voltage (small), and W the angular frequency of the signal. For the greatest sensitivity the a.c. amplitude is 38

measured at a fixed time in the life of a drop, but the mean value is easier to measure for instrumental reasons. The a.c. amplitude 75,76 goes symmetrically through a maximum at a summit potential Es equal to El, and has a maximum value given by :

2 2 1 a = -n F AVc(WD)2/4RT (13) s

As can be seen, both eqns (12) and (13) are linear in concentration. They have not been averaged over the drop time because in their derivation from the diffusion equations, allowance was not made for the expanding d.m.e., in contrast to the Ilkovic equation.

The residual current is 11/2 out of phase with the alternating voltage75 whereas, for a fast reaction at low frequencies, the faradaic current is 17/4 out of phase. Thus the residual current can be eliminated by measuring the alternating current in phase with the alternating voltage, which will give only a faradaic component. This is achieved by incorporating a phase-selective device into the circuit, such as a mechanical interrupter.

The analytical possibilities of a.c. polarography are75,76 that processes with E only 40 mV apart are separable, instead of 200 mV s with d.c. polarography. The effects of preceding faradaic processes are therefore greatly reduced, which makes it possible to determine small amounts of an ion in a large excess of a more noble ion. Moreover, measurement of a peak enhances accuracy, but the major advantage of a.c. polarography is that the effect of the residual current cannot be nearly so effectively suppressed with d.c. polarography. This latter fact is the one which makes a.c. polarography significantly more sensitive to traces of metals than its d.c. counterpart.

5.2 EXPERIMENTAL PROCEDURE

The platinum substrates were foils (2.54 cm square by 25 pm thick) and were welded onto platinum wire (460 pm diameter), sealed into pyrex glass tubing. The platinum wires were welded to copper leads. These 2 foils (13 cm geometric area) were cleaned40 prior to platinization by . dipping them in boiling dilute aqua regia, washing, dipping in warm concentrated nitric acid, washing, cathodizing in 0.1M sulphuric acid -2 at 30 mA cm for 10 minutes, and finally washing again. 39

Low-lead deposits were prepared from a solution of 3.5% (w/v) chloroplatinic acid and 0.005% lead acetate in 2M hydrochloric acid -2 at 30 mA cm for 10 minutes. A single-compartment cell was used with two silver-silver chloride counter electrodes in parallel, one on each side of the platinum foil. Moderate stirring was maintained, by means of a magnetic stirrer, and there was no gas evolution from any of the electrodes. (With slow stirring gas was evolved from the platinum cathode.) The constant current was obtained from a type L30 stabilised voltage/current supply unit manufactured by Farnell Instruments Ltd.(Sandbeck Way, Wetherby, Yorkshire). Deposition potential was measured, via a Luggin capillary containing the plating solution, versus a Radiometer K401 saturated calomel electrode with 12 a Radiometer pH meter 4. This had an input impedance of 10 0 and -12 a grid current of less than 3 x 10A. The apparatus is shown schematically in Figure 3, and the variation of deposition potential with time of plating is shown in Figure 4. These values fluctuated by approximately t 5 mV. The deposits were black and of a velvet 2 looking texture. A preliminary experiment with a 10 cm substrate gave 92 mg of deposit, which corresponds to 100% coulombic efficienty. However, the potential of deposition for this electrode was about 70 mV more cathodic than for the others. There is no obvious explanation for this, other than perhaps that the different geometry of the electrode provided different stirring conditions, and hence a different potential at the same current density.

High-lead deposits were prepared from a solution of 3.5% chloroplatinic acid and 0.2% lead acetate in 0.2M hydrochloric acid at 10 mA cm-2 for 30 minutes. A two-compartment cell divided by a sintered glass disc was used with a platinum counter electrode. The working electrode 0 was turned through 180 (without being removed from the plating solution) after 15 minutes of platinization. Moderate stirring was maintained in the working compartment and tnere was no gas evolution at the electrode being platinized. Chlorine was evolved from the counter electrode. The power was supplied and the deposition potential was measured as before. The apparatus is shown schematically in Figure 5 and the variation of deposition potential with time of plating is shown in Figure 6. These values fluctuated by. approximately 4,- 3 mV. The deposits were grey and 40 Figure 3. Apparatus for preparing a low-lead Platinized Platinum Electrode.

Stobihsed Power Supply

- A31A90. Pt --113 )A30.

Plating Solub on

marstic St;rrer

Counter Electrode Construcbion /1,9 wire was prot-ect-ecl Sy Araldibe . czakeSive to Sop it From beIns eQUer, away durirts L'iori •

A3lAt Figure 4. Variation of Deposition Potential with Time, for Platinization from a 3.5% Chloroplatinic Acid plus 0.005% Lead Acetate Solution, 2H in Hydrochloric Acid -2 at 30 mA cm

10 Time iminuas 42

Figure 5. Apparatus for preparing a High-Lead Platinized Platinum Electrode.

Stabilised Radiometer pH Meter 4 Power Supply

SCE-0

Sotubion _

• — di Ma r,eEic Stirrer Sinteredt Glass Disc Porosity i Figure 6. Variation of Deposition Potential with Time for Platinization from a 3.5% Chloroplatinic Acid plus 0.2% Lead Acetate Solution, 0.2M in -2 Hydrochloric Acid at 10 mA cm .

.1-

10 • 20 30 Tirne minutes 2 smooth. A 13 cm substrate gave 70 mg of deposit, which corresponds to 6o5, coulombic efficiency.

To test the electrochemical reversibility of the electrodes, the e.m.f.s of the cells

SCE I Pb(NO3)2 I Pb-Pt

were measured with a Radiometer 4 pH meter. The Pb-Pt electrodes were washed with water for 3-4 hours after platinization, and were then placed in 0.1M, 0.01M and 0.001M lead nitrate solutions and the potentials measured over periods of several hours. Some of the electrodes were previously anodized at 1.255 V (SCE) in 1M perchloric acid solution. The reference electrode SCE was a Radiometer K401 saturated calomel electrode, separated from the lead nitrate solution by a 30% potassium nitrate and 3% agar salt bridge.

Prior to the kinetic leaching experiments the electrodes were soaked in water, which was periodically changed, to remove lead acetate occluded from the plating solution. The soaking times were 60 hours for the low-lead deposits and a minimum of 30 hours for the high-lead deposits. The actual leaching runs were carried out by immersing the electrodes in 130 cm3 of leaching solution, contained in a pyrex vessel with a slowly rotating magnetic stirrer and fitted with a perspex lid. Periodically the whole of the solution was replaced with fresh leaching solution, and analysed. One set of runs was carried out in 1M perchloric acid at open circuit potential, which was monitored as for, the e.m,f. experiments except that the salt bridge was 30% ammonium nitrate and 3% agar. Similar runs were performed in 1M perchloric acid with the platinized platinum electrode held at 0.855, 0.955 and 1,255 V(SCE). The potential was monitored as before but using a Luggin capillary, and was kept at the value desired with a platinum foil counter electrode immersed in the leaching solution. In case any lead had plated out on the counter electrode, the latter was at first washed after each run in boiling 1:1 nitric acid and the resulting solution analysed for lead. As the analysis was always negative, this procedure was discontinued.

In most experiments the leaching solution was 1M perchloric acid, open to the air. Since no lead leaching was detected from low-lead deposits, additional experiments were confined to the high-lead ones. 45

To test the effect of pH, one run was carried out instead with 1M sodium perchlorate solution, exposed to the air. A final experiment, designed to test whether oxygen played a role in the dissolution, was carried out with 1M perchloric acid purged with hydrogen. The latter was purified by passing it through a Johnson Matthey Al silver-palladium membrane diffusion unit. To remove surface oxygen from the platinized platinum electrode before the experiment, it was left in water overnight with hydrogen bubbling through it. The gas also passed through a flask containing, concentratedperchloric acid (an amount calculated to give 1M perchloric acid when mixed with the water in the cell) which had been put in series with the reaction cell. The run was started by blowing the perchloric acid from the flask into the reaction cell with nitrogen. After 4 hours the solution was replaced by aerated 1M perchloric acid and oxygen was bubbled through it.

All the above experiments were carried out at room temperature (23 + 1°C).

The chloroplatinic acid was supplied by B.D.H. The perchloric acid was B.D.H. Aristar, and the sodium perchlorate, lead acetate, lead nitrate and nitric acid were of AnalaR quality. All water was doubly distilled, the second time from alkaline permanganate solution. The gases used were supplied by B.O.C.

The lead ions leached out were analysed by a.c. polarography, which was sensitive to ca. 2 x 10 7 M lead. The cell was of the Heyrovsky type and thermostatted at 25°C. Before each analysis, the system was purged with nitrogen for about 5 mins. A Cambridge General- Purpose Polarograph and Univector Polarograph Unit were used with a Bryans Autoplotter 22000 Series X-Y recorder, measuring the mean a.c. amplitude. The capillary constants measured at the potential of the SCE in 0.1M potassium chloride solution were m = 2.60 mg s-1, t = 2.97 s, at h = 72.9 cm. For each series of analyses the polarogram of unused leaching solution was first recorded, and then the polarogram of this with a standard addition of lead nitrate solution. The lead ion concentration in the samples was then determined by comparison. A typical series of a.c. polarograms is shown in Figure 7. An initial analysis was carried out by d.c. polarograpny, and this is shown in Figure 8. • As can be seen, the sensitivity is much lower than that for a.c. polarography.

Figure 7. A Typical Series of A.C. Polarograms.for Lead Determination, from a High Lead Platinized Platinum Electrode.

10.6 cm 5.15)(1076 NI Pb2+ Calibration j1j11

111

10.6 cm

10 ml oF M ,A4\00011,4! 5 F.41 of 1.03 x10-z M jnkValhilf1111 16 Tv 4604,1 11 4 111 H C104 1/11111 1A1 4114.11 1,14 4i PO r 1144 11111011 4V4 111 Wireil /410 101 Alf 1 Sensitivity .V9 <---Electrocie Potential Figure 7 continued

Estimation

1.8 crn Si 0.67 x10-6 PI P62.4 3.5 orn 1.70)(10" P61+ 6.0 om 2.92x10"6 1'1 PO* o1 pb in 130 mi. .%4-3 fay of Pb in 130 ml I'. 74, 1.49 of Pb in 130 mt.

aifti 4 „ 161141.

Semple 3 (3.00 kr) Sample 2. (1.00110 Sorrip. le 1 (0.17 kr) Figure 7 continued

Estimation

0.5 cm 0.2.4-x10-6 M Pb2A. 0.5 cm 0.2.4.x t0-6 IA PO 0.8 CM F..: 04 3c1X10-6 M * 61,4-3 0P Pb ih 130.011 .6 bts of Pb in 130 ml of Pb in 134 mL

Ka11 I, ldt~i~1 -14 1 vri 0.5 cm ' 0.8 cry ri

Scvnilte 6 (24-.00 Inr) Snmete 02.25 hr.) Sample 4 (6.00hr)

' o .1111111. ...p..... 111111111,1 .., " .. = • N -- 0- ...__Q-_....-- _0 c -- -I- C..> 0 0 0 _ . . .- --•-. t " •' ---.-_ -- . ---..-. -CI- 4,t " rff flit ,=, ,, .,.. „,,,ts-tzr;;.• ' -.1fEl•• !,,0,1”'' ,JJ to -4. •-, ..., r !„d •1012 - _....,,,:f.Itt " _ „ .1.L .... •,iiigiil • . r71,..• • Lot in 1:1 r--1- ic' (.1...,.'`! ' • , .. ',i• ' ill MI "1" IF 1.11.01 ,,Fi7;s1.' Allipii ts II' 14 -I , 01.1 • . ---cc- w , • • ., ..y.,•0'i,..,.1,si'-' -74:, • -111:7 ,,-.•• 1.2 . -.-- 7--'.,: •-s.,•" "- . ,..,. "I• ''' '..:.-,•:4iiip1/1il:477-TlihD - - ,* .,4 ,..;;.44.4 wolvy • . . ' '-'' ,p 01 ,0!..iitfiliiitti MIMI .4, ,iiiii, " ' ;'•:. .r.:,'• .0...Tr cji44' ';4 Mill , . ..ii . pr...(4.4.., , _ - ,--_ „4:, 4 ,...... IIMIni C> 'I; i ' 0 4•.., - I ••••••- 1 EMI. ____0 ..„.‘-7.-?24 I' mom .., . f• i _IL ., . ,A * .1-:,-- ....4t, ,---

11111Er - -- -1 rtllj 1111111111111111=1111 IM11111111111111111111 .t • I=MIIIII 111111111111111111111111111 II/ 4, IIIIIII MEM i r i cs r g EME 1. , 1 MEI — • 511111111111 c) MC= (1,1 ENNIO

[ 4 : [ 1 / • (E

71 41 0 UR V`i V .I--_ -,"1 0 1 D 11 W1;3°— _0101:1141 cp : si 1 t 1 L 4DVIaq ;o uo _art -3.'gi 0113.-20-;--surrta o-arTozi - I _i _.1 ,.. L : o 1 __:___....._1 R , , 1 , •i i ; -,-- , ...... , I1 , i , _ 50

5.3 RESULTS

In the e.m.f. experiments the potentials of both the low- and high-lead electrodes drifted between 570 and 600 mV(SCE) with time and with changing lead nitrate concentration9 often in the reverse direction to that predicted by the Nernst equation. The potentials of electrodes that had been anodized generally drifted downwards from roughly 800 mV(SCE), irrespective of the concentration of the lead nitrate solution.

Leaching experiments with the low-lead electrode produced no detectable lead ions (i.e. < 2 x 10-7M) after immersion in 1M perchloric acid for 24 hours. The potential of the electrode decreased from around 850 to 820 mV(SCE) during the run.

The results of the leaching experiments with the high-lead electrodes are presented in Table 5. Initial experiments snowed that the nigh-lead plating solution, on repeated use, turned progressively 2- from yellow to red as PtC14 ions accumulated. Moreover, at a given stage, the deposits obtained suddenly changed from being grey and smooth to being grey and fairly rough, and the deposition potential simultaneously dropped from about +300 to +250 mV(SCE). The last two observations are consistent25 with depletion of lead acetate in the plating solution, and (for lead acetate concentrations above 0.01%) 46 this should produce some increase in the area of the resulting deposit. The rough deposits yielded much more lead than the smooth ones, typically 370 pg as compared with 160 1..6. This large increase in the amount of lead leached out cannot be accounted for fully by the expected increase in area. Structural changes in the deposit may have taken place as a consequence of the altered composition of the plating solution, in 2- which, as already stated, a nigh concentration of PtC14 would have built up. In subsequent experiments, plating solutions were only used while the deposition potential remained at ca. 300 mV (SCE) at 10 mA cm-2, and only smooth deposits were accepted.

The time dependence of lead leaching in 1M perchloric acid from the high-lead electrode is shown in Figure 9; the half time is approximately 15 minutes. The dissolution did not fit zero, first, or second-order kinetics. The potential of the electrode decreased from about 830 to 770 mV(SCE) during the run. Anodization of the platinum electrode at 0.855 and 0.955 V(SCE) in 1M perchloric acid

Table 5

Data on the leaching of lead from a high-lead platinized platinum electrode in 1M perchloric acid

Preparation Leaching

V/mV(SCE) Appearance E/m V(SCE) time/hr Mass Pb/µg Remarks

288 to slightly 832 to 0.48 287 Potential 274 rough 765 .34.90 485 floating 256 to slightly 794 12.00 343 Potential

270 rough 955 18.05 0 floating for 22.13 0 first 12 hr then fixed

250 to slightly 6.00 371 Potential 254 rough floating 12.00 576 floating all 24.00 381 the time

235 to fairly 798 to 1.17 433 Potential 247 rough 808 3.17 529 floating all 6.12 6o4 the time 12.12 677

250 to fairly 6.25 206 Potential

216 rough 1255 12.17 218 fixed-but not 24.17 228 reached for 30.92 247 0.17 hr after start

241 to rough No leaching experiments were done with 228 these rough electrodes 206 to rough 199

294 to smooth 0.17 74 Potential 317 1.00 117 floating all floating 3.00 139 the time 6.00 149 12.25 155 24.00 161

Table 5 (continued)

Preparation Leaching

V/MV(SCE) Appearance E/m V(SCE) time/hr Mass Pb/µg Remarks

304 to smooth 0.18 44 Potential 297 1.00 0 fixed - but 3.00 0 not reached 1255 6.00 0 for 0.08 12.00 0 in. after 24.00 0 start 35.00 0

314 to smooth 518 0.17 0 Leaching 303 to 1.08 0 solution was 482 3.00 0 1M NaC104 ; 6.00 0 potential 12.00 0 floating 24.00 0 35.00 0

270 to smooth -252 to 4.00 0 With H2. 278 -249 purge for 4 hr 740 to 7.50 51 With 02 purge 820 for 305 hr

262 to smooth 855 1.72 Potential 290 1255 7.47 118 fixed at 2 values, analysed at end.

245 to smooth 955 10.47 158 Potential fixed 289 0.18 hr after start

285 to smooth Electrodes used for 291 e.m.f. runs 276 to smooth 306 53

Figure 9. The Time dependence of Lead Leaching in 1M Perchloric Acid from a High-Lead Electrode.

150

Z. "rab 100 U a) 1 .0 GL 0 50

0 5 10 15 20

Time of Liaa.c6in3/ 6otArs

0 Ftoating Potential

0 Potential held at 855ta(SCO For First 1:72.hours and then at 1255 mV(SCE) For c3. Further 7.47 haorS

Potent a(, hetd at '155 mV(SCE) For 10,47 hours 54

did not affect the lead leaching, but at 1.255 V(SCE) no lead dissolved. Note, from Table 5, that anodization of a rough electrode at 1.255 V(SCE) -did not completely prevent lead from dissolving, but only suppressed it by a factor of nearly 3. This also could presumably be due to structural changes in the deposit, as for the above area changes.

Lead leaching did not occur when the high-lead platinum electrode was immersed in 1M sodium perchiorate over 35 hours. The potential of the electrode decreased from about 520 to 500 mV(SCE) during this run. Nor did any lead dissolve when a high lead platinum electodel pre-treated with hydrogen gas, was soaked for 4 hours in 1M percnloric acid purged with hydrogen. The potential of this electrode increased from -252 to -249 mV(SCE) during tne run, which was the potential expected for a hydrogen electrode. After 4 hours the electrode was removed to a fresh sample of 1M perchloric acid through which oxygen was bubbling. The potential rose to 740 mV(SCE)'after 15 minutes and to 820 mV(SCE) after 3.5 hours, and in this time 51 .ig of lead was leached out.

5.4 DISCUSSION

The non-Nernstian.response to the electrodes makes it impossible to assign any thermodynamic significance to the e.m.f.s measured in 72 this paper. Those in the literature refer to cast alloys and should be similarly investigated. Nonetheless, the very fact that the electrodes were electrochemically irreversible while the Pb++ I Pb-Hg system is highly reversible77 indicates a very low lead activity or, to put it another way, strong binding in the deposit. Determination of lead activity coefficients by vapour pressure measurements should prove interesting.

The kinetic experiments with the high-lead electrode snow that both oxygen and acid Must be present before lead ions are produced, and it seems a logical step to ascribe the dissolution process to reaction (9). A closer examination of all the evidence, however, indicates otherwise. In the first place, the initial open circuit potential of dissolution is only about 0.133 V less than the Nernst potential of the oxygen electrode when allowance is made for activity coefficients and the large (15 mV) Henderson liquid junction potential. From known electrode kinetic data,78 a cathodic overpotential of this 55

magnitude would, at the electrode in question, reduce oxygen to water at a rate of 0.14 x 10-9 equiv. s . (If the liquid junction potential is in fact very much less than the Henderson value, as the experiment with the hydrogen electrode suggests, the rate is increased to 0.19 x 10-9 1 equiv. s'.) Lead, however, is dissolving at the rate of 1.2 x 10-9 equiv. -1 s . An analogue of reaction (9), in which oxygen is reduced to H202 instead of to water, is ruled out by the low standard potential of the H202/02 couple. One is tnerefore forced to conclude that some lead in the platinum lattice is already present in the oxidized form, presumably as PbO. A second piece of evidence is provided by the anodization experiments. Lead was leached at the same rate at 0.855 and 0.955 V(SCE) as at open circuit potential (0.83 - 0.77 V(SCE)) but ceased at 1.255 V(SCE). This potential, which is 1.515 V(NHE) when allowance is made for the liquid junction potential, is just above the standard potential of the reaction

+ Pb02 + 4H 2e Pb244 2H2 0, E° = 1.455 V (14)

2+ If Pb ions are present in the deposit, they would be oxidised to Pb02 at this potential and would not appear in solution..

In the experiment with the hydrogen purge the Pb0 will have been reduced to lead, and it is interesting to note that on subsequent passage of oxygen both the rate of lead dissolution and the initial open circuit potential were relatively low. The rate of oxygen reduction at this potential was probably high enough to account for the slow rate of lead oxidation, and equation (9) might hold. The relatively small total amount of lead leached out (51 pg compared with the expected 140 pg after 3.5 hours) may be explained by the slow rate of diffusion of oxygen into the platinum matrix to form more PbO.

The amount of lead leached out in 1M HC104 in the presence of air reached a limit of approximately 161 pg. If the lead content of 46 the deposit is taken to be 2.7%, the 70 mg deposit should have contained a total of 1.91 mg lead. Thus only 8.4% of the total lead dissolved. Tne high lead plating procedure used snould give an electrode of roughness 46 2 factor 296 and thus an area of 0.38 m ; since the density of platinum is 21.45 g cm-3, the deposit should be approximately 8.5 nm thick. This corresponds to roughly 30 atomic layers, taking the interatomic distance 56

in platinum crystals79 to be 0.2775 nm. Thus, if the lead is assumed to be homogeneously distributed throughout the deposit, the results indicate that the lead from only the first 2 or 3 atomic layers is dissolved. Further lead dissolution would be limited by its exceedingly slow diffusion through the platinum crystal lattice and by the slow diffusion of oxygen into it.

A high-lead platinized platinum electrode of the kind described will produce an 8 pm solution of Pb when immersed in 100 cm3 of acid solution. This is not likely to prove harmful except in research demanding a high level of purity, as in conductance measurements with dilute acid solutions, or in work in which lead could subsequently act as an electrode poison. Even in these cases, the experiments described above show that the surface lead can be effectively removed by first soaking the electrode in 1M perchloric acid solution for a few hours. 57

PART III

THE SURFACE AREA OF PLATINIZED PLATINUM ELECTRODES 58

CHAPTER 6

THE MEASUREMENT OF ELECTRODE SURFACE AREA

The prime aim of platinization is to obtain a large area. The concept of true area on a microscopic scale is somewhat arbitrary, and especially so with electrochemical systems. Thus it has been 80 argueu, that the area determined in the dry state, e.g. by the BET method, may be much greater than that associated with electrochemical processes, because the inside surfaces of many pores and cavities of a highly porous surface are not effective in, or contribute but slightly to, the transfer of charge in an electrochemical process. Since all investigations of the areas of platinized platinum electrodes have used electrochemical methods of estimation, the principles of these methods will be outlined first. As will be seen, the correlation between gas adsorption and electrochemical methods of area determination is reasonable for platinized platinum electrodes.

6.1 DEFINITION OF USEFUL PARAMETERS

The results of area measurements have unfortunately been presented in the literature in a wide variety of ways, and we must begin by defining the many terms used. The simplest measure is the surface area itself, S. Tne area lacks significance on its own and is never used in tnis tnesis. For electrodes, the most useful quantity is roughness factor 0. It is dimensionless and defined by

115= S/A ( 1 5 ) where A is the geometric area of the substrate electrode. The specific area is also used, although this is generally more useful for powders than electrodes. Two types of specific area nave been employed, and are described by different names in this review to avoid ambiguity. First, the mass specific area, a , is given by Ttl 0 = S/m ( 1 6 ) • ,m 59

where m is the mass of the deposit. Note that this does not include the mass of the underlying substrate. Second, coulombic specific area, G , defined by c

= (17) c where q is the amount of electricity passed in deposition. It is necessary to use ac because some investigators have characterized the amount of deposit only by recording the number of coulombs passed during platinization, and 100% coulombic efficiency cannot be assumed. For the deposition of platinum with complete coulombic efficiency, a -1 mass specific area of 1978 cm2 g is equivalent to a coulombic specific 2 -1 area of 1 cm C . Division of the roughness factor by the amount of deposit, or of specific area by geometric area, gives the corresponding -1 -1 specific roughness factor, in g or C . There is no advantage in using this quantity and it will not be employed here.

Another useful quantity is the degree of platinization of electrodes. Like specific area, it is quoted in two forms. The mass degree of platinization, W , is defined as m W = m/A (18) m and the coulombic degree of platinization, W, is defined by

W = q/A (19) c For 100% coulombic efficiency, a mass degree of platinization of -2 1 g cm is equivalent to a coulombic degree of platinization of -2 1978 C cm . It follows from the above that

= S/A =GLD =GW (20) m m c c

6.2 ELECTROCHEMICAL DETERMINATION

There are three basic electrochemical methods for estimating the surface area of platinum electrodes: the determination of the amount of electricity equivalent to the formation of a monolayer of hydrogen, that equivalent to a monolayer of oxygen, and the measurement of the double layer differential capacitance. Each of these three methods refers to a fairly well defined region of potential of the platinum electrode in some inert electrolyte solution (usually 6o

sulphuric or perchloric acid). The solution is saturateuA39a with an inert gas, such as N2 or He, so that the system is not complicated by the ionization of adsorbed hydrogen or the reduction of adsorbed oxygen.

6.2.1 THE THREE POTENTIAL REGIONS OF PLATINUM

The nature of the platinum-solution interface is conveniently illustrated by use of galvanostatic charging curves, as shown in Figure 10. The electrode, under the inert conditions previously mentioned, is polarized with a constant current and the change of 39a,81,82 potential with time is followed. Anodization from the potential of hydrogen evolution gives a slow increase in potential as adsorbed hydrogen is ionized to approximately 0.35 V(RHE). Then a sharp rise occurs while most of the charge passed goes into charging the double layer, to approximately 0.8 V(RHE). Finally, there is a slow linear increase, during which a layer of adsorbed oxygen is formed, to oxygen evolution at about 1.6 V(RHE). Subsequent 39a cathodization removed the adsorbed oxygen; as this process involves a high activation energy the potential drops sharply to an almost horizontal region at about 0.7 V(RHE). The potential then decreases further to the hydrogen evolution region. The flatter sections up to gas evolution are known as arrests.

A more recent technique is linear sweep voltammetry (Figure 11), where the potential of the electrode is varied linearly with time and the current is monitored. In fact, this gives82 '83 a differential plot of Figure 10, actually in the reciprocal form. If the galvanostatic charging curve is plotted as E against t, the differential (slope)is dE/dt, and dividing this by t he constant current (dQ/dt) gives dE/dQ. With the voltammogram, the varying current (d(Vdt) can be divided by the constant sweep rate (dE/dt) to give dQ/dE as a function of t. With linear sweep voltammetry, approximately the same three potential regions as for the galvanostatic charging curve can be discerned. Below ' 0.4 V(RHE) the hydrogen adsorption peaks occur at the same potential on forward and reverse sweep, indicating the reversibility of the hydrogen adsorption reaction. 39a,82 During the anodic sweep the double layer charging region stretches 61

Figure 10. Schematic charging Curve for Platinized Platinum in an Inert Electrolyte.

1.5.

0.5 •••••••=6

Anodic C aLko

Potentials (E) are versus tne reversible hydrogen electrode (RHE) for the solution used. UnderAalvanostatic conditions, the charge passed (Q) can be' replaced by the time .of charging. 62

Figure 11. Schematic Linear Sweep Voltammogram for Platinized Platinum in' an Inert ,Electrolyte.

C.)

0 C

0.5 1.0 E/mV(R1-10 63

from about 0.4 to 0.85 V(RITE) and the oxygen region then extends up to 1.4 VCRHE), above which oxygen is evolved. On subsequent cathodiz- 39a,82 ing the oxygen reduction peak is again displaced towards less noble potentials because of its irreversibility.

The most recent techniques applied to investigating the platinum- solution interface are optical; viz. ellipsometry and reflectance spectroscopy, where39a,84,85 the change in the polarization state of the light reflected from the platinum surface is recorded. The principles of the physical optics of the two techniques are85 identical, which differ merely in the methods of observing the change in polarization 39a,84-87 of the reflected light. They confirm both qualitatively and quantitatively the physical significance of the three potential regions of platinum.

There is general agreement on the nature of the platinum surface at potentials in the hydrogen region and in the double layer charging 81'82'85-87 region. The former involves two types of reversibly bound adsorbed hydrogen atoms, while the latter consists of39a,81,82,85-87 the bare platinum-solution interface. But there is substantial contention on the nature of the platinum surface in the oxygen region. 88 For this, tnere are four different postulates:

(a) Oxygen is chemisorbed and the surface can accommodate more than one oxygen atom per surface platinum atom.

(b) Oxygen is chemisorbed but the amount in excess of one atom per surface platinum atom is sorbed into a surface region a few atomic layers deep. This later type of sorption was suggested by Warner and Schuldiner89 and called dermasorption.

(d) Distinct platinum oxides of various stoichiometries exist for all potentials. 82 As Gilman has pointed out, such criteria are somewhat arbitrary and the properties of the anodic film on platinum do not allow an 82 unequivocal distinction to be made. Moreover, he considers such classification as not worthwhile, but appears not to favour (b). 64

84 Hoare,39a on the other hand, favours (b), while Damjanovic supports 88 (c). Postulate (a) is the most recent and due to Biegler and Woods, whereas (d) seems not to be advocated by anybody nowadays.

Recent work on the oxygen region can be divided into electrochemical or optical studies. The electrochemical workers used chronopotentio- metry (galvanostatic charging curves) and linear sweep voltammetry. 83 88 One group ' found that on polarising platinum in sulphuric acid to extreme anodic potentials (3.0 V(RHE); greater than any one else), the oxygen in the surface layer reached a constant level at 2.2 V(RHE) which was independent of time of polarisation. They concluded that 90 oxygen was chemisorbed with a maximum 0:Pt ratio of two. Others have found completely opposite behaviour for polarisation up to 1.8 V(RHE) and they favour dermasorption. Work on the dissolution and diffusion m91 of oxygen into anodically polarised platinu tends also to favour dermasorption.

The optical investigations used ellipsometry and reflectance spectroscopy. To calculate optical constants for the system these 85-87,92-95 workers have assumed that a surface oxide film was formed. However, for consistent results, at 1.5 V(RHE) a film thickness of at 85-87,92 least 0.5 nm is necessary and this is too large for the oxide 85-87,92-95 film to be a monolayer. Also, there is definitely a profound change in optical properties at approximately 1 V(RHE). It seems95 that the oxygen is chemisorbed up to 1 V(RHE) when it reaches monolayer coverage, and phase oxides develop th4reafter with platinum in both the oxidation states of 2 and 4. The distinction between a thin phase oxide and dermasorbed oxygen is tenuous, so the weight of the evidence would appear to favour a combination of postulates (b) and (c).

6.2.2 EMPIRICAL DETAILS OF AREA DETERMINATION

Use of the hydrogen region has been most popular for determining areas, and galvanostatic charging appears to be favoured up to now (Figure 10). If, on anodizing, the potential is not allowed to exceed about 0.8 V(RHE), the subsequent cathodic charging curve is identical 13,96 to the anodic one and there is no hysteresis. This fact is made use of in area determinations using the hydrogen arrest. The electricity 65

81 equivalent to a monolayer of hydrogen is found from the length of the hydrogen arrest in the charging curve. Monolayer coverage is 81,82 usually assumed to occur at 0 V(RHE), although this is not certain. The necessary correction for double layer charging can be simply achieved by extrapolating the linear double layer portion to 0 V(RHE). The charging curve can be determined either anodically or catnodically (see above), and it was claimed by the originators97 of this technique that, provided the solution was free from oxygen and "negative metal impurities", these were the same. They also recommended that the current density should be high enough to make leakage currents negligible. However, on the wnole two different schools of measurement have evolved and will be dealt with separately.

Anodic measurement of the hydrogen arrest was developed in Russia and has been thoroughly investigated for platinized platinum and platinum black. The measurements were carried out galvanostatically 13 -2 and were best at moderate current densities (ca. 0.1 mA cm ) where little or no hysteresis was found and the system was virtually in equilibrium96 all the time. Comparison of the areas of various electrodes obtained by the krypton BET method with the lengths of the anodically determined hydrogen arrests has shown that the latter -2 98 require 280 pc cm for platinized platinum in 0.5M sulphuric acid and for platinum blacks99 in 0.05M sulphuric acid, and 218 pC cm-2 for platinized platinum in 1M HC1 and in 1M KBr plus 0.03M HC198 and for 63 platinum blacks in 0.5M sulphuric acid. The accuracy of these measurements was reckoneuA98 to be + 16%. Nevertheless, it has been 100 argued that the anodic measurement of the hydrogen arrest is unsatisfactory as a routine determination because allowance has to be made for the molecular hydrogen generated at the reversible hydrogen potential, 101,10a as was found for a Teflon-bonded platinum black electrode. The results in fact depended on the starting potential. It must be noted 81 that in most of the Russian charging curve work the potential did not go below approximately 10 mV(RHE), so no molecular hydrogen would have been formed. However, in none of the experimental area calibrations 63,98,99 was the lower limit of potential given.

The cathodic measurement of the hydrogen arrest was developed chiefly in the U.S.A. but has not been experimentally tested like the 66

47,100,103 anodic measurement. Galvanostatic charging at high current -2% densities (ca. 100 mA cm '), where the areas were independent of the current, have been used. Assuming that at the reversible hydrogen potential there was one atom of hydrogen per surface platinum atom, 100 -2 it was calculated that 210 kC cm were required. With 1M perchloric acid or 0.5M sulphuric acid as electrolyte, this figure has been used 4 47 for smooth7100,103/ and for platinized platinum.

Linear sweep voltammetry is another, more recent, technique for examining the hydrogen region (Figure 11). The hydrogen adsorption is reversible, hence the areas under the cathodic and anodic waves are 82 equal, and they are proportional to the area of the electrode. An approximate correction for double layer charging is simply achieved82 by using as a base line for integration, lines tangential to the minima at about 0.5 V(RHE). The cathodic limit for integration is arbitrary but as for galvanostatic charging could be set as.0 V(RHE). In practice, the most cathodic minimum in Figure 11 has been proposed82 as the lower limit. As shown for a cathodic sweep in Figure 12a, 82 this choice implies that any charge required to complete the hydrogen monolayer at further cathodic potentials is compensated by a contribution from hydrogen evolution at more anodic potentials, area abd is equal to area bed). Note however that the extrapolations used in Figure 12a to separate the hydrogen adsorption and evolution currents do not fulfil.„ 83 the requirement that the sum of the partial currents should equal the experimental current; those in Figure 12b satisfy 83 this constraint. Recent work questions this division of the current. Comparison of voltammograms for smooth and platinized platinum in IM sulphuric acid showed the minimum to be at 80 mV(RHE) for smooth 83 and at 40 mV(RHE) for platinized platinum. This shift was attributed to the mass transport control of the hydrogen evolution reaction, diffusion of molecular hydrogen away from the interface being restricted by the porous nature of the electrode. It was concluded that integration using the most cathodic minimum as the lower boundary, gave a charge for smooth platinum corresponding to a coverage of 0.77. Experimentally, linear sweep voltammetry with the conversion factor -2 661l0* of 210 pc cm for the hydrogen region has been used for determining the areas of smooth platinum electrodes. 67

Figure 12. End Point for Monolayer Hydrogen Coverage in a Cathodic Linear Sweep Voltammogram of Platinized Platinum in an Inert Electrolyte. (a) E /mV(R1-1E) 0 0.2. 0.4

(b) E im\ARH 01 0 4.

---Measured current —DoubLe teayer chcwgin9 current: ----Hypobhebicat po.rbfal ciAv-rei* or ikevabioh 68

The analogous use of the oxygen arrests is not so straight- 90 forward as for the hydrogen ones since a condition of full coverage by oxygen is difficult to define. This is because surface oxides of differing thickness and/or formal valencies are produced, as discussed in the previous section.

The correction for double layer charging in the oxygen anodic arrest can be carried out by extrapolating the steep double layer line to the potential just before oxygen evolution commences. For the cathodic arrest this correction must be calculated from capacitance data, but is very small because the potential range covered is short. Measure- 39a ment under galvanostatic conditions was satisfactory, although in one 100 case roughening of the electrode resulted (see Chapter 3.1+). A fairly high current density was always necessary39a otherwise the lengths of the cathodic and anodic arrests differed. During the anodization there is• a danger of oxygen dissolving in the platinum, but the effect of this 39a can be circumvented by making fast cathodic measurements only. To reduce this danger even further, some workers nave not cathodized from 105 101 102 oxygen evolution but from 1.21+ V, 1.10 V(RHE), and 1.05 V(RHE); in the first case the electrolyte was 1M sulphuric acid at 25°C, so the potential corresponds to ca. 1.10 V(RHE), and for the other two cases o 105 the electrolyte was 85% orthophosphoric acid at 150 C. It was found that for cathodization from 1,24 V in 1M sulphuric acid to be consistent 2 with krypton BET measurements, 272 pc cm was required for platinized -2 platinum, as opposed to 31+7 pc cm by comparison with smooth platinum, -2 and 355 pc cm for Teflon-bonded platinum black electrodes. A very -2 102 similar figure of 357 pc cm was obtained for Teflon-bonded platinum black electrodes in 85% orthophosphoric acid at 150°C when cathodization was started at 1.05 V(RHE). Some workers nave used values calculated from the packing density of platinum by assuming that a complete monolayer of oxygen is formed with a 1:1 0:Pt stoi- L -2 46 -2 chiometryi givingl06 1+20 pc cm and. 513 pc cm according to the packing densities used. In the latter case the charge equivalent to 46 oxygen coverage was determined by linear sweep voltammetry.

Finally, the double layer differential capacitance as a measure of surface area has been investigated in some detail. The classical 69

a.c. bridge technique is not suitable for determining the capacitances of platinized platinum electrodes because their areas, and thus their capacitances are too large. In essence, the method has been to take the slope of the charging curve in the (linear) double layer region. The differential capacitance, C, is then given by Lt C = f I dt/tE (21) 0 where I is the current which causes a change in potential of LIE in a time At. Usually galvanostatic conditions13,97 have been employed so the slope could be measured directly and eqn (22) used.

I (22)

9 Ershler recommended that moderate current densities be used otherwise the system is not in equilibrium. If the potential is changed too quickly some of the current is still used for the hydrogen arrest (in the case of anodic charging) or the oxygen arrest (cathodic charging). 107,108 In a later modification applied to platinized platinum, the starting potential was fixed by auxiliary d.c. polarization and a square wave signal of low amplitude (S 20 mV) and frequency (2 Hz) was applied. The change with time of the potential across the test electrode was recorded on a cathode ray oscilloscope. In more recent 106 work, short constant current pulses of less than 10 psec duration are employed so as to eliminate any faradaic contribution. 46161,1o8 Whilst the differential capacitance method is good for 46 61 98 smooth platinum, it is not usually satisfactory ' ' for finely divided platinized platinum (although it has been successful on some 44 107 occasions ' ), nor for Teflon-bonded platinum black fuel cell 101 electrodes and platinum black.63'99 Its usefulness appears to 46 be limited by the texture of the electrode, and forms of high 63 roughness factor and specific area may give anomalous results because the values of the capacitance in narrow pores are too high compared with those of a smooth surface. For platinized platinum the best figures, from comparison with the krypton BET method, are: -2 -2 36 5 pF cm in 0.5M sulphuric acid,98 20 .4- 3 pF cm in 1M HC1 .70

or 1M KBr plus 0.03M HC1,98 and 20.1 4. 0.7 pFcm-2 in IM sodium 107 sulphate solution. The experimental conditions pertaining to the last figure were dubious. Measurements were taken using a square wave of frequency 2 Hz and of amplitude 20 mV centered on 445 mV or about 0.86 V(RHE). Since this is a little above the upper limit of the double layer region, oxygen adsorption would affect the value -2 of capacitance measured. Values of about 20 1.1F cm have been 82 criticized for smooth platinum as being too low, and attributed to contamination. CHAPTER 7

CO-ORDINATION OF PAST AREA WORK

7.1 DEPENDENCE ON PLATING CONDITIONS

The roughness factors reported in the literature vary over an enormous range. Extreme values as high as 3900 or even 20000 were 107 said to be obtained by a "standard" platinization procedure. 44,46,105 However, usual values are 200-500, although roughness ,44,47 44 factors in excess of 2000 have been claimed in one instance with the proviso that this was true only with large deposits (approx. 2 1 g of deposit per cm of electrode). The actual value obtained depends on the amount of deposit, the composition of the plating solution, the current density of deposition, and the temperature. The first three have been investigated at ambient temper'atures only, but temperature itself seems to nave been somewhat neglected. Moreover, experiments designed to test the effect of the composition of the plating solution have all in fact just examined the effect of the lead acetate concentration.

7.1.1 DEGREE OF PLATINIZATION 13 3657 The areas of platinized platinum electrodes increased ' with degree of platinization, although the rate of this increase 13,36, usually dropped 44 as the platinization continued, or even attained a maximum value. The available data are plotted in Figures 13, 14 and 15. 13 Figure 13 is based on the measurements of Slygin and Frumkin. 2 They plated a platinum substrate electrode of geometric area 34 cm -2 from a 0.041M (2%) chloroplatinic acid solution at 2.9 mA cm . Electrode surface areas were given in coulombs, obtained from the anodically measured hydrogen arrest in 0.5M sulphuric acid, and were 2 961- converted to cm by dividing by 280 IX cm `. Tne original and derived data are presented in Table 6. 72

Figure 13. Variation df roughness factor 0 (0) and mass specific area m 0:0 of a platinized platinum deposit with mass

degreeofplatinizationulm•Obtained-- from a 0.041M (2%) chloroplatinic acid solution at 209 mA cm-2; from Ref•13. 2000 30

0 i 500_ 6 0

10 easszemes.

5-00 20

2 0 4-0 60 m /t-n9 73

Table 6

Data concerning the variation of roughness factor and mass specific area with mass degree of platinization

2 For a platinized platinum deposit (A = 34 cm ) obtained -2 from a 0.041M chloroplatinic acid solution at 2.9 mA cm (Ref.13). See text and Figure 13.

Data given in Ref.13 Derived data

Deposit Hydrogen Hydrogen arrest/cg-1 wini/Mg cm-2 ,e5 a Am2 m mass/g arrest/C Deposit mass -1 mg

0.030 0.63 21 0.88 66 75 0.226 4.5 19.9 6.6 470 71 0.723 9.8 13.6 21.2 1030 48 1.135 15.1 13.3 33.4 1590 47 2.5 19.6 7.8 73.5 2060 28. 74

Figure 14. Variation of roughness factor gr (0) and coulombic specific area oo (0 of a platinized platinum deposit with coulombic degree of platinization U. Obtained from a 0.006M (0.39) chloroplatinic acid plus b.6x10-4M (0.025%) lead acetate solution, 0.025M in hydrochloric acid, at 0.7 mA cm-2 for 10 minutes with current reversal every 10s1--from - Rpf;7;

•••••••••••1•10 30

•••••11MMIMMIRr 20

0 G 0 0 20 40 60 Nc C cm 2 75

Table 7

Data concerning the variation of roughness factor and coulombic specific area with coulombic degree of platinization

For a platinized platinum deposit obtained from a 0.006M chloroplatinic acid plus 6.6 x 10-4M lead acetate solution, -2 0.025M in hydrochloric acid, at 0.7 mA cm with current reversal every 10 s (Ref.7). See text and Figure 14.

Data given in Ref.7 ' Derived data

Degree of Polarization WIC cm-2 14 0 Am2 C-1 platinization/ capacitance/pF C C (electrode)-1

0 36.5 0 . (0.26) - 3 819 0.2 6 28 6 1380 . 0.4 lo 23 12 ' 2230 0.8 16 19 42 4020 3.0 28 10 90 5670 - 6.4 40 6 45o 13000 31.8 92 3 800 7900 56.6 56 1 Figure 15. Variation of roughness factor (Pof a platinized platinum deposit with coulombic degree of platinization W. Obtained from a 0.05M (2.4%) chloroplatinic acid plus 2.7x10-4m (0.01%) lead acetate solution at a series of constant potentials; from Ref.57.

• 195 tyN • 220 mV

• 0 246- hIV 12 295" m\/ •- 345 vr‘V

20 40 coc / C 77

Table 8

Data concerning the variation of roughness factor with coulombic degree of platinization at different constant deposition potentials

For a platinized platinum electrode obtained from a 0.051 chloroplatinic acid plus 2.7 x 10-4m lead acetate solution at five different but constant deposition potentials (Ref.57). See text and Figure 15.

Data plotted in Ref.57 Derived data

-2 Deposition potential C cm Capacitance/ 4 mF/ cm-2

-50 mV(SCE) 12 42 2100 [195 mV(NHE)] 22 76 3800 45 154 7700

3 12 600 12 46 2300 -25 mV(SCE) 24 84 4200 [220 mV(NHE)] 41 146 . 7300 47 172 8600 55 200 10000

10 36 1800 0 mV(SCE) 20 70 3500 [245 mV(NHE)] 35 124 6200 46 164 8200

4 11 550 11 38 1900 +50 mV(SCE) 16 54 2700 [295 mV(NHE)] 23 76 3800 41 136 6800 53 170 8500

4 ' 11 550 6 18 900 +100 mV(SCE) 15 44 2200 [345 mV(NRE)]. 26 72 3600 26 74 3700 38 110 5500 40 116 5800 78

Figure 14 is based on the capacitance measurements of Jones and Bollinger.7 They plated two circular platinum electrodes of 3 cm diameter and 0.85 cm apart in a conductance cell, from a solution — 0.006M (0.3%) in chloroplatinic acid plus 6.6 x 10 (0.025%) in lead -2 acetate and 0.025M in hydrochloric acid, at 0.7 mA cm with current reversal every 10 s. The experimental data were polarization capacitances, at 700 - 3070 Hz, measured in the conductance cell with the plating solution as electrolyte. The parallel plate capacitance resulting from the two electrodes was negligible (ca. 60 pF) compared to the polarization capacitances (30 - 8000 µF) with which it was in 109 110 parallel. To give areas, values obtained at 1 kHz were selected. 107,9/8s These were doubled (two electrodes in series) and divided by -2 20 pF cm to give areas. Roughness factors were obtained by using 2 a geometric area of 14.14 cm . This gives 0 = 0.26 for the unplatinized electrode, and a better area measure might therefore be obtained by multiplying all values by four. The original and derived data are presented in Table 7.

Figure 15 is a modified version of one due to Bernard.57 The plating was carried out from a 0.051 (2.4%) chloroplatinic acid plus -4 2.7 x 10 m (0.01%) lead acetate solution at a series of constant potentials. The expet'imental data were farads per geometric square centimeter measured in 0.03M potasium chloride solution, and were -2 'converted to roughness factors by dividing107798'; by 20 µF cm . The original and derived data are presented in Table 8.

The roughness factors of Figure 15 are almost 10 times greater than those in Figure 13 and these, in turn, are some 10 times greater than the ones in Figure 14. The flf values in the last case might be 109 expected to be well below the true values, partly because of a faradaic contribution to the measured capacitance, and partly because , of the very small current density (0.7 mA cm-2)'of platinization. This would normally lead to a low coulombic efficiency and the 2 accumulation of PtC1 in the plating solution, except that here the chloroplatinic acid concentration was correspondingly small. It is therefore a moot point whether the coulombic degree of platinization here is a fair guide to the mass degree of platinization. The exceedingly high roughness factors of Figure 15 may not be very reliable 79

111 as they are based on double layer differential capacitance measurements wnith, as already pointed.out (Chapter 6.2.2) can be anomalously. nigh.

Not only the magnitudes but also the shapes of the plots in the three diagrams differ markedly. The maximum in the roughness factor curve in Figure 14 is not encountered elsewhere, and the complete linearity shown in Figure 15 might be a consequence of the potentiostatic, as opposed to galvanostatic, plating conditions. A more likely explanation is proposed below. In a more recent and quantitative study,3b in which electrode area was measured by anodization in the hydrogen region, eqn.(23)

16 = bt ( e- t) (23) was found to relate roughness factor to deposition time, t, at 100 mA -2 cm for the first 400 s of plating from a 0.041M-(2%) chloroplatinic -1 acid solution. The values of b and S varied from about 0.3 to 1.3 s -1 and from about 0.0006 to 0.003 s respectively, depending on the pretreatment of the platinum substrate. For platinization from a 0.1M (4.9%) chloroplatinic acid plus 2 x 10-3m (0.076%) lead acetate -2 solution at 100 mA cm the area increased linearly with time (as in Figure 15) or with mass of deposit. From the figures given the present author calculates that after t s :

Id - (3.8 x 10-3)t (24)

by dividing98 the rate of area increase, found by the hydrogen arrest -2 methods by 280 pC cm . The pretreatment of the substrate had no effect in this case.

At 10 MA cm-2, with the same lead-containing solution, the relationship between area and time or mass of deposit was no longer linear. The deposition rate varied with electrode pretreatment and the rate of increase in surface area decreased with deposition time, as in Figures 13 and 14. The experimental data followed the equation

0m = g (i/t)Y (25) 8o

• where i is the current density of 10 mA cm-2, and g varies from about . 650 to 850 and y from about 0.3 to 0.4, depending on the nature of the substrate surface. Equations (24) and (25) taken together suggest that, at high current densities specific area is approximately constant with time of plating (or degree of platinization) whereas, at low current densities specific area decreases with time of plating. This is confirmed by Figures 13-15, for the potentials of deposition of Figure 15 are of a magnitude to give current densities higher than those of Figures 13 and 14.

As expected, this general rise in area with degree of platinization (with the exception of Figure 14) is reflected in a continued enhancement of electrode activity as more platinum is deposited. An example 112 is found in the electro-oxidation of methanol.

7.1.2 LEAD ACETATE CONCENTRATION

The concentration of lead acetate used in the plating solution affects the roughness factor of the final deposit. The mass specific -2 areas of deposits3b prepared at 100 mA cm from 0.1M (4.9%) chloroplatinic acid solutions decreased as the lead acetate concentration was raised from 10 4M (0.0038%) to 0.05M (1.9%). The magnitude and type 46 of decrease was not specified. On the other hand, Thacker, who had used only 10 mA cm-2, found that the area, measured by the oxygen arrest technique, was a maximum at a Pb/Pt atomic ratio of 0.0021 which corresponds to a lead acetate concentration of 8 x 10-5M (0.003 %). 2 This is shown in Table 3 and Figure 16. The data in terms of cm - 2 were divided by the geometric area of 0.16 cm to give roughness factors. Rather more indirect evidence on the area-lead content relationship is 12 provided by the measurements of Hevesy and Somiya. From their cathodic polarization data the present writer has estimated a measure of the surface area and these results are plotted in Figure 17. 12 Hevesy and Somiya platinized platinum from a 0.062M (3%) chloroplatinic acid solution, 0.2M in hydrochloric acid, with lead acetate added to the extent of 0 - 0.0029M (0.11%). The current 2 -2 density was stated to be 10 cm probably a misprint for 10 mA cm . A measure of area has been derived from their cathodic overpotential 81

Figure 16. Variation of roughness factor 0 of a platinized platinum deposit with lead content (0) and with the lead acetate concentration of the plating solution (D). Obtained with a 0.05lM (2.5%) chloroplatinic acid solution at 10 mA cm-2 for 1 hour with current reversal every 1.5 minutes; from Ref.460 Lead Acetate Conce.ntrabion/103M --

,Q1 0.02

Pb/Pb. Atomic Ratio inDeposib 82

Figure 17. Variation of the "area" of a platinized platinum deposit with lead content and with the lead acetate concentration of tne plating solution. Obtained from a 0.062N (3%) chloroplatinic acid solution 0.214 in HC1. The current density was stated to. be 10 cm2 , probably a misprint for 10 mA cm-21 from Rfe .12 .• Lead Acebate Concenbrabion/1011

I 2

0.02 0,04- 0,06' Pb/P1 Atomic Rctbio iv\ DeFosib 83

Table 9

Data concerning the variation of "Area" with lead content and lead acetate concentration

For a platinized platihum deposit obtained from a 0.062M cnloroplatinic acid solution 0.2M in hydrochloric acid at • 10 mA cm-2 (Ref.12). See text and Figure 17.

Data given in Ref.12 Derived data r Lead acetate Lead content r/MV at Lead acetate Pb/Pt "Area" conc./% of deposit/% 20 mA conc./10'M atomic I/-2 cm-2 ratio in A cm deposit V-1

0 0 - 103.7 0 0 0.193 0.00045 0.035 97.9 0.012 0.00033 0.204 0.0023 0.15 81.0 0.060 0.0014 0.247 0.023 1.5 77.5 0.60 0.014 0.258 0.111 7.1 84.4 2.93 0.067 0.237 84

measurements in 0.5M sulphuric acid at a constant current density -2 of 20 mA cm of geometric area. Current (I) versus overpotential (11) curves are functions of the exchange current density (4) or the exchange current (10 = 10). At constant current, 71 increases as 10 decreases, and (if io is constant) the smaller 10 , the smaller is the area S. In the special case when overpotentials are less than approximately 20 mV

I/71 = = Io(nFART) = io S(nE/\11T) (26)

where n is the number of electrons involved in the reduction and V the stoichiometric number. Thus, at constant current and small overpotentials:

cc I/T1 (27)

-1 -2 In this way measures of "area", in A V cm of geometric area, have been deduced. The original and derived data are shown in Table 9 .

Figure 17 displays an area maximum at a Pb/Pt atomic ratio of approximately 0.005 - 0.01 which corresponds to a lead acetate concentration of ca. 2.5 X 10-4M (0.01%), but there is no minimum at higher lead acetate concentrations as there is in Figure 16. This 12 46 work and Thacker' s taken together indicate that the area would be maximized by employing a lead acetate concentration of about -4 1.3 x 10 m (0.005%). These two pieces of work are augmented by 112 a study on the variation of electrode activity, to methanol oxidation in 2M Me0H plus 3M H2 Soy,solution at 25°C, with the lead acetate concentration of the plating solution. For a 0.103M (5%) plating solution of chloroplatinic acid and deposition at a constant potential of 250 mV/ a maximum activity of the electrode occurred for a lead acetate concentration of 5.3 x 10-4m (0.02%). However, at deposition potentials of +50 mY or -50 mV, monotonic increases in electrode activity with lead acetate concentration were found up to 4 x 10-3m (0.15%). Unfortunately this work is not strictly comparable with that discussed above, partly because a different property was measured and partly because of the potentiostatic conditions of preparation. The potentials 85

employed probably correspond to current densities rather larger than -2 46 12 the 10 mA cm used by Tha:cker and by Hevesy and Somiya.

It should be added that both the area and the activity measurements were carried out in acid solution. Introduction of oxygen at any stage would have resulted in some leaching of the lead from the surface layer and a possible concomitant change in area (see Chapter 5).

7.1.3 CURRENT DENSITY OF DEPOSITION

When the solution composition, as well as the mass of deposit, are kept constant, the area depends upon the current density employed. This is shown by the work of Slygin and Frumkini3 in Figure 18. Electrode surface areas were given in coulombs, obtained from the anodically measured hydrogen arrest in 0.5M sulphuric acid, and were 2 -2 converted to cm by dividing by 280 pc cm . The original and derived data are presented in Table 10. The type of information 44 conveyed in Figure 19, due to Joncick and Hackerman, though super- ficially similar, differs because the masses of deposit ranged from 1 to 23 g, and the specific area varies with the mass of deposit (cf. Figures 13 and 14). A third piece of information can be deduced 64 from a set of current-voltage curves for platinized platinum electrodes carried out in 1M hydrochloric acid under 1 atmosphere of hydrogen. The composition of the plating solution was not stated. According to the argument used to derive Figure 17, the initial slopes of current versus voltage should be approximately proportional to area. Estimation of the slopes is made difficult by the small size of the 64 graph on which the results are plotted, but the diagram does show that the area tended to increase with current density of platinization -2 up to a value of 45 mA cm . There is no evidence of any maximum. 36 In the study of Roth and Lasko on the effect of current density of platinization (i) on area, using 0.1M (4.9%) chloroplatinic acid plus 2 x 10-3M (0.076%) lead acetate solution, the results were presented in two forms. For a constant quantity of electricity during platinization, PS was plotted against i. The form of this plot was essentially the same for all platinum substrates in spite of various pretreatments (see Chapter 3.2.1), and showed a maximum -2 roughness factor of about 200 at i = 20 mA cm (using the conversion 86

Figure 18. Dependence of the mass specific area am of a platinized platinum deposit on the current density of platinization i. Obtained front a 0.041M (2%) chloroplatinic acid solution, giving a constant mass of deposit of 0.719; from Ref.13.

5 rt.% A crrt-2. 87

Table 10

Data concerning the dependence of mass specific area on the current density of platinization

2 For a platinized platinum deposit (A = 34 cm ) obtained from a 0.041M chloroplatinic acid solution, giving a constant mass of deposit of 0.71 g (Ref.13): hence, -2 = 20.9 mg cm . See text and Figure 18. Wm

r Data given in Ref. 13 Derived data 2 Current Hydrogen Hydrogen arrestAg-1 i3 Cm/cm Deposit mass -1 density/ arrest/6 mg mA cm-2

0.58 16.4 23.1 1720 82 2.9 - 9.45 13.3 993 48 14 5.55 7.8 583 28 88

Figure 19. Dependence of the mass specific area Cm of a platinized platinum deposit on the current density of platinization 1. Obtained froma.... 0.205M (10.0;61. 0) and a 0.090M (4.4% E cbloroplatinic acid solntion.O.86M in Hdlg giving depoSiis . of masses ffom 1 to 23g; from Ref.44.

100 200 . 300 400 ilmA cm-2. 89

factor employed for eqn,(24)). Since the coulombic efficiency varies with current density, the results cannot be directly related to those of Figures 18 and 19. However, the results were fitted36 in terms of mass specific area to the equation

Cm = ht (i/t)C (28) 01, where it = constant. From their logarithmic plot of 0;11/t against C is close to 0.5. It follows that C is almost independent m of current density.

These four area-current density determinations seem united only 13 by their diversity. In one case the mass specific area decreases 44 monotonically as i increases, in another it increases monotonically, 44 -2 in a third it passes through a maximum at 250 mA cm , and in yet 36 another it is altogether independent of current'density. Of the 64 two studies of roughness factor, one snows fo increasing continuously 36 -2 with i and the other records a maximum in kf at 20 mA cm . One's scientific morale is sustained only by the thought that these differences may have arisen as a result of the diversity of plating conditions, and by the hope that further, carefully planned, studies may disentangle the variables. It is strongly recommended that, in future, workers always describe their surfaces in terms of roughness factors as well as specific areas.

7.1.4 POTENTIAL OF REPOSITION

Some complementary evidence is available from plating experiments carried out under potentiostatic conditions. The dependence of roughness factor on the (constant) potentials57 of platinization, shown in Figure 15, points to a maximum roughness factor at a potential of approximately 230 mV. This is illustrated more clearly in Figure 20, in which the slopes of the lines in Figure 15 (i.e. the coulombic specific areas) have been plotted against the potentials of platinization. The data are given in Table 11. Since lower (more cathodic) plating potentials correspond to nigher current densities, it follows from Figure 20 that the roughness factor should pass through a maximum with increasing current density. The decrease in coulombic specific

Figure 20. Dependence of the coulombic specific area ab of a platinized platinum deposit on the deposition potential of platinization. The plating solution had the same composition as for Figure 15, and the coulombic specific areas were calculated from the slopes of the lines of - that figure; from Ref.57.

180

r4E

160

14-0 1 200 250 300 350 Deposition Pobentict.L m\/

Table 11

Data concerning the variation of coulombic specific area with deposition potential

For a platinized platinum electrode obtained from a 0.05M chloroplatinic acid plus 2.7 x 10-4M lead acetate solution at constant deposition potential (Ref.57). The coulombic specific areas were calculated from the slopes of the lines of Figure 15; see text and Figure 20.

Deposition potential /cm2 C-1 mV(SCE) mV(NHE

- 50 195 172 - 25 •220 184 0 245 178 50 295 164 +100 345 144 92

area snown at deposition potentials more anodic than 230 my may be explained as being.due to a decrease in coulombic efficiency as 2- PtC14 is produced and accumulates. The researches described below confirm that, at potentials up to +230 mV, higher areas are obtained at more positive potentials. 113 In one work in which deposits were prepared from 0.041M (2%) cnloroplatinic acid solution 1.5M in sulphuric acid, the mass specific area (measured by the hydrogen arrest technique) rose from 2 -1 40 - 70 cm mg when deposited between -50 and +50 mV(RHE), to 2 -1 170 - 200 cm mg when deposited at potentials above +150 mV(RHE) as shown in Figure 21. The activities of these electrodes to the electro-oxidation of methanol increased by 8-10 times when the deposition potential was changed from -50 to +250 mV(RHE), and most of this change occurred for potentials between -50 and +50 mV(RHE). -2 However, the mass degree of platinization ranged from 5 - 15 mg cm and this detracts from the usefulness of the above observations. 112 A similar trend was found for electrodes prepared from a 0.103M -h (5%) chloropiatinic acid plus 5.3 x 10 ,i (0.02%) lead acetate solution, but here most of the change in activity occurred between +50 and +250ffiV. 61 In a fourth paper, the electrodeposition from a 0.051M (2.5%) chloroplatinic acid solution 1M in HC1 was studied at the controlled potentials 0, 50 and 150 my; both the areas determined by the hydrogen arrest technique and by differential capacitance at 0.6V showed a very large increase for a given mass of deposit when the deposition potential was changed from 0 to 50 mV, and this was confirmed visually from 6l electron micrographs. X-ray diffraction line broadening indicated particle diameters of 24 and 15 nm respectively for these two deposition potentials, and it was concluded that at the lower potential a high coverage of hydrogen on the platinum surface inhibited the nucleation process, with a consequent increase in particle size and a smaller surface area.

It is easier to explain the phenomena observed during electrodeposition if constant potential rather tnan constant current is employed. This is done sin the research described in Chapter 8. But however the plating is carried out, whether potentiostatically or galvanostatically, both potentials and current densities should 93 Figure 21. Dependence of Mass Specific Area and Coulombic Efficiency on Deposition Potential; from Ref.113.

100

2.00

:5 150

Cra E

50 0

'100 200 V/ rnV(R1-1E)

0 2. % chLoroplobinic aci4 (From Pt meto() +1.5 M su(plzie aced 0 2 % chLoropla.tinic occd (FreshIcommercia0-1-1.5- M stAphunqc Acad 0 2°7o chroropta &ince cicici(commerant) 0 2.. 10 chloroptatinic cid(useci,conlmerciathi.5-11.Tulphuric odd 94

be reported. Although the value of the uncontrolled variable will change somewhat during the course of plating, an average value, with the range of variation if it is significantly large, would provide valuable information for subsequent workers.

7.1.5 TEMPERATURE

The effect of the temperature of deposition on the area of platinized platinum electrodes has received but little attention. 113 Electrodes prepared from a 0.041M (2%) chloroplatinic acid solution 1.5M in sulphuric acid at -20, +4, and 4-80°C displayed the same type of dependence of activity (to the electro-oxidation of methanol) on deposition potential as did that prepared at +20°C. Deposits obtained at 200 - 250 mV(RHE) (180 - 230 mV) in the temperature range -20 to +80°C had similar values of mass specific area and . practically the same activity despite a thirty-fold variation in deposition currents. Since mass specific area depends on current density as well as on degree of platinization, these two factors appear to have compensated one another.

7.2 REPRODUCIBILITY

In the literature there are widely differing reports about the reproducibility of platinized platinum electrodes. The data in Figure 22, for example, show a reproducibility of 5 - 10)%. Electrodes 114 prepared from 0.103M (5%) chloroplatinic acid plus 5.3 x 10 m -2 (0.02%) lead acetate solution at 120 mA cm gave activities for the T1 - Ti "electron exchange catalysis which were reproducible to 5%. Even better consistency was obtained53 in work on the anodic polarography of dextrose; rotating disc electrodes prepared from 0.062M (3%) chloroplatinic acid plus 1.6 x 10 3M (0.06%) lead acetate -2 at 152 mA cm and 1300 r.p.m. gave polarographic current reproducible to 2%. In contradistinction, Joncich and Hackerman44 found that platinization was only "reproducible" when no lead acetate was added. Using a solution 0.049M (2.4%) in chloroplatinic acid and 2.5 x 10 5M -2 (0.001%) in lead acetate,. and 210 mA cm , they obtained samples of 2 -1 specific areas 8.9 and 15 cm mg after 30 minutes of plating and 2 -1 samples of specific areas 2.9 and 6.8 cm mg after 120 minutes. 95

A solution of 0.09M (4.4%) chloroplatinic acid containing no lead -2 acetate, with a current density of 150 mA cm , gave samples of 2 -1 specific areas 1.5 and 2.0 cm mg (or with roughness factors of 350 and 620 respectively); the time of plating was not given. Thus, even these "reproducible" results are not very promising,

although the reproducibility of two points in Figure 19 is good to 5%.

7.3 DECREASE WITH .TIME

It is highly desirable for the area not to change with time. This will be so for surfaces which are not far from thermodynamic equilibrium or which have a low surface self-diffusion coefficient, 115 D. Its value for platinum, over all orientations, was found from the interference microscopic observation of the decay of sets of parallel scratches on platinum crystal surfaces, and obeyed eqn.(29) between 890 and 1310°C

D = D (29) s o exp(-0t/RT)

-5 -1 -1 where D = 4 x 10 cm2 s and R = 108 10 kJ mol . The coefficients o were strongly orientation dependent. The Arrhenius equation was also 116 verified by field-electron emission microscopy in the temperature -1 range 823 to 1123°C, and the activation energy of 123 4. 13 kJ mol is in quite good agreement with the previous value. If eqn.(29) -22 2 -1 applies down to room temperature, D at 25° is 4.7.x 10 cm s This corresponds to a mean diffusion distance for a single atom of -11 -6 3 x 10 cm per s or 2.6 x 10 cm per day. Whether tnere is any net movement of material in a given direction will depend on the existence of a gradient of chemical potential on or up to the surface. Such a gradient might well be present on a freshly prepared platinized platinum deposit, and appreciable sintering would then be expected. For the evidence we must turn to the literature. 13 An electrode freshly prepared from a 0.041M (2%) chloroplatinic acid solution decreased in area by about 30% in 2 - 3 months, which 117 averages out as 0.1% per day. The changes of area with time of three identically prepared platinized platinum electrodes in 5M orthophosphoric acid at 25 and 80°C are shown in Figure 22. The decrease Figure 22. The decrease in area with time of platinized platinum electrodes. Theyowere immersed in 5M orthophosphoric acid, maintained at 25 for the first 12 days, and at 80 for 7 hours. per day on 13 of the remaining 22 days. All three electrodes (0 0 t) were prepared from a 0.051M (2.5%) chloroplatinic acid plus 8 x 10-5N (0.003%) lead acetate solution, at 10mA cm-2 for 1 hour with current reversal every 1.5 minutes; from Ilef.117.

0 90 30 Time clays 97

in area is only about 0.8% per day at 25°C but becomes ten times as large when the temperature is kept at 80°C for 7 hours a day. A 102 Teflon-bonded platinum black electrode lost 4.4% of its area per day when stored in 85% orthophosphoric acid at 150°C.

These long-term linear decay results stand in contrast with other observations which point to rapid initial decay which levels off. A platinized platinum electrode, prepared as described in Figure 15 but with the potential of deposition not mentioned, was periodically immersed57 in 0.05M KC1 solution and the double layer differential capacitance measured after 10 minutes and again after 18 hours. Between each such pair of measurements the electrode was stored in -air in an anti-dust container for many weeks. After only 1 day of ageing the differential capacitance decreased by 20% during the i8 hours of immersion, but the subsequent rate of decrease was only 13% over the next b months, or an average of 0.07% per day. It is not clear whether the initial drop was caused by a genuine contraction in area or to a process of equilibration (perhaps involving adsorption) at the interface. Deactivation by contamination is also a distinct possibility, although Bernard57 discounted this on the grounds that the solutions were very pure. Evidence for incidental contamination 66 comes from the observation that severe poisoning of a smooth platinum electrode occurred when it was left in 'pure" 1M sulphuric acid for long periods (hours) at open circuit or held at potentials below about 0.9V. Also, the mass specific area of precipitated platinum black63 found from double layer capacitance in 0.5M sulphuric acid, decreased sharply, while the (lower) value determined by benzene adsorption changed only slightly. The period of time involved and the storage environment were not stated. Further support for the initially rapid type of decay comes from catalysis studies. On a platinized platinum rotating disc electrode (of unspecified preparation) the cathodic hydrogen diffusion current i was measuredll8 and found to vary with the rotation speed W according to

1 = a + -••• (30) VW

The existence of the intercept a was ascribed to the fact that the 98

active site density p on the surface was not infinite and, from a, values of p were derived. They were found to decrease from 11 -2 10 -2 >10 cm immediately after platinization to 1.3 x 10 cm six -2 days later and 0.6 x 10/0 cm after twelve days. In fuel cell work,112 electrodes deposited from 0.103M (5%) chloroplatinic acid -h plus 5.3 x 10 14 (0.02%) lead acetate solution, displayed activities to methanol oxidation which stabilized in approximately 3 to 6 weeks to values which were only 20 to 30% of the initial activities. Perhaps the essence of this type of decay lies in the interaction 119 of the surface with gases. It was noted long ago that prolonged exposure of platinized platinum to air destroyed its activity for 6 use as a hydrogen electrode and also reduced its wettability. The type of gas is important too: the area (as measured by the hydrogen arrest) of a platinized platinum electrode (of unspecified preparation) 120 decreased in 2 hours by 8% when maintained at 100°C in air and by o 503; at 300 Ci but sintering at 300° in hydrogen for 2 hours diminished the area by a factor of thirty.

The decrease in area is more rapid if the electrodes are kept in a polarized state. Thus, the roughness factor of a platinum electrode, platinized from a 0.041M (2%) chloroplatinic acid solution 47 1M in hydrochloric acid at 50 mV, dropped from 2380 to 97 in 100 hours when kept at 0.6 V in 0.5M sulphuric acid plus 1M acetic acid at 94°C. /13 Similarly, an electrode that was being anodically polarized at 5 mA o cm at 00 C, and which had been plated from a 0.041M (2%) chloroplatinic acid solution 1.5M in sulphuric acid at 20°C, lost 69% of its initial surface in 48 hours. However, the area lost was only 28% in 48 nours if the electrode had been prepared instead at 80°C (Table 12). 113 n 0 It was suggested that more regular crystals had been formed at 00 C and that these recrystallised less on standing.

Brightp.plated electrodes, obtained by using chloroplatinous acid, 50 were initially active as hydrogen electrodes but lost their activity rather rapidly, especially when stored in hydrogen. 99

Table 12

Dependence of the decrease with time of mass specific area on the temperature of deposition

The platinized platinum electrode was prepared from a 0.041M chloroplatinic acid solution 1.5M in sulphuric acid at +250 mV(RHE). During the ageing it was anodically polarized -2 o at 5 mA cm at 80 C (Ref.113).

/ 2 -1 Time after Mass specific area/cm mg plating/hours Prepared at +20 C Prepared at +80C C

0 175 181 24 66 48 54 130 88 120 100

CHAPTER 8

TILE VARIATION OF AREA AND COULOMDIC EFFICIENCY OF PLATINIZED PLATINUM ELECTRODES WITH DEPOSITION POTENTIAL AT CONSTANT MASS DEGREE OF PLATINIZATION

The major aim of platinizing platinum is to obtain a large surfade area. To this end optimum plating conditions must be sought. Yet (as was emphasized in Chapter 7.1.3) there is extensive disagreement between different workers on the way in which the area of platinized platinum electrodes varies with the current density (and hence the potential) of deposition. One reason for this disagreement lies in the diversity of area measures, and the three most important ones were defined in Chapter 6.1. It was pointed out that both mass and coulombic specific areas sometimes varied with degree of platinization. Thus, the differences reported in the literature are attributable to the measure of area used, the electrochemical plating conditions, and the fact that in all studies, except one, the mass degree of platinization varied from run to run.

Deposition at constant potential is theoretically more meaningful than at constant current. There have only been two previous such 113 studies, namely that of Podlovchenko and Petukhova and that of BernaruA57 (see Chapter 7.1.4). Together these two studies cover deposition potentials from -42 to +345 mY. However, there seemed to be a requirement for a fully integrated study over as wide a potential range as feasible (-50 to +594 mV, almost twice as large), at constant mass degree of platinization.

The study also enabled coulombic efficiencies to be measured, compared with literature data, and interpreted theoretically.

8.1 EXI'ERIMENTAL PROCEDURE 2 The platinum foil substrates (A = 13 cm ) were welded onto platinum wire, which was sealed into the soda glass end of graded 101

seal glass tubes. The substrates were cleaned prior to platinization by dipping them in boiling dilute aqua regia, washing, dipping in warm concentrated nitric acid and washing again. They were dried at 110°C for 30 min, weighed, treated once more with warm concentrated nitric -2 acid, washed, cathodized in 0.1M sulphuric acid at 30 mA cm for 10 min, and finally washed.

For most runs platinization was carried out in 50 ml of an aqueous solution of 0.072M (3.5% 1,4/v) chloroplatinic acid plus 4 1.3 x 10 m 0.00%) lead acetate at 25.0 + 0.1°C. For comparison, three additional experiments were performed with different plating solution compositions. Two were without the addition of lead acetate and one was with the plating solution made 2M in hydrochloric acid. Each plating solution was used once only. The plating cell was divided into three compartments by sintered glass discs of porosity 1, with a platinum foil counter electrode in each of'the two end compartments, as shown in Figure 23. The working electrode section was stirred fairly vigorously with a magnetic stirrer. The deposition potential was maintained constant to better than 1% with. a TR70/2A potentiostat (Chemical Electronics Co., Birtley, Durham, England) and was monitored on a Radiometer pH meter 4. The reference electrode was quinhydrone in 0.144M hydrochloric acid. It was connected to the working electrode via a Luggin capillary containing the same acid. This hydrogen ion concentration, equal to that in the plating solution assuming complete acid dissociation of H2PtC16 , was chosen so as to give a low liquid junction potential. A value of 0.3 mV (with the plating solution negative w.r.t. the Luggin capillary) was calculated from the Henderson equation and ,the following limiting molar conductances: 140.4 cm2 n-1 mo1-1 for19 PtCd; 76.4 for121a Cl and 349.8 for121a 11+. For the plating solution which was 2M in hydrochloric acid, the reference electrode was quinhydrone in 211 HC1 solution.

After platinization each electrode was stored in 1M perchloric acid for about 24 hours before the area was determined, to leach out the surface lead from the deposit (see Chapter 5). This was necessary since an electrochemical method was used to determine the area, and lead ions in 'solution could have interfered with the system. 102

Figure 23, Plating Apparatus

Pobentiostab TR70/2A

Radiometer pH Meter 14.

0.144M .1-1ydrochlork Acid

---

AN. TI,errnosb-at Levet

Porosiby 3 Pt 4 Porory I Plating Sol til:ion Saw-cited of Sintered Glass Discs QUinhydrone in 0.144.M Hrlrockloric Acid Stirrer 103

To supplement the coulombic efficiency data it would have been • 2- 2- useful to measure directly the concentrations of PtC16 and PtC1 after platinization. The only feasible laboratory technique available was spectropnotometry. To this end the visible/u.v. spectra of potassium hexachloroplatinate(IV) in water and potassium tetrachloro- platinate(II) in 2M hydrochloric acid (to suppress hydrolysis) were recorded, using a Hitachi Perkin-Elmer model 124 dual beam spectrophotometer with a model 165 recorder. The results are summarised in Figure 24 and the peaks listed with literature values in Table 13. Surprisingly, the spectra do not show a wide divergence at any wave- length in the range covered, despite the substantial difference in 2- 211 colour of the two species - PtC16 solutions being yellow and PtCl ones red. Presumably the origin of the colour differences lies at higher wavelengths. Consequently, spectrophotometry was not suitable for determining the species present in the plating solution, and the idea was abandoned.

The real surface area of each electfode was determined from the length of the hydrogen arrest of the galvanostatic charging curve, taken in 1M perchloric acid at 25.0 + 0.1°C. The Luggin capillary contained the same solution, and quinhydrone in 1M perchloric acid acted as reference electrode. A two-compartment cell, divided by a sintered glass disc of porosity 3, was used with a platinum foil counter electrode, as shown in Figure 25. Both compartments were sealed from the atmosphere and continually purged with oxygen-free nitrogen (DOC) which had been scrubbed with 1M sodium hydroxide solution and then with 1M perchloric acid. The nitrogen was initially bubbled through the perchloric acid in both compartments for 1.5 hours via the inlet tubes. The platinized electrode was then electro- 82 chemically cleaned as follows: 2 s at 1.8 V(RHE) with stirring; 30 s at 1.2 V(IIIIE) with stirring followed by 90 s without stirring; 10 s at 0.4 V(RHE) without stirring. The stirring was effected by the bubbling nitrogen, and it was stopped by turning the double-oblique tap so that the gas passed instead over the surface of the perchloric acid in the working electrode compartment. The nitrogen bubbling was then resumed for a further 30 minutes. Platinized platinum 104 Figure 24. Visible and ultra violet spectra of potassium nexachloroplatinate(IV) in water and potassium tetracnloroplatinate(II) in 2M hydrochloric acid

a 0 _1

+-K2 ncit,. in zM HCl

K PtC.16 ipwctber 200 400 600 A/nrn 105

Table 13

The visible/u.v. absorption peaks of aqueous nexacnloro- platinate(IV) and tetracbloroplatinate(II) solutions

1. 2 PtC16 N2 Ptc14

M /nm Ref. ri /nm max max max max Ref. IC 451 44.1 473 15.5 Na 440 60.3 122 K 470 16.6 122 NEt6 453 50 123 476 15.0 125

K 350 435 389 59.7 Na 348 603 122 392 52.5 122 NEty 353 490' 123 It 392 59 125 K 390 58 b 126 IC 261 16900 323 70.0 Na 264 14100 122 328 81.3 122 Na 265 26920 a 124 331 64 125 NEt4 262 24500 123 330 69 b 126 IC 260 721 K 259 631 122 264 250 125 IC 260 794 b 126

K 216 9630 217 9580 125 a In 2M tlydrocnloric acid solution b Read from a graph 106 Figure 25. Area Determination Gloss Screw Apparatus Thread Toints

Sintered Gloss Disc Porosity 3

Thermostat Level

IN Perchl °Pit Acid

Saburoted Solution of Quirthydrone it 1M Perchloric Acid

SIDE ELEVATION . Sintered Gloss Dist Porosity 3

Nitrogen Outlet System

water

Nibro9en Inlet <—"System

END ELEVATION 107

electrodes thus treated should retain their clean (or "active") 82 state for hours. Witn'the solution quiescent, galvanostatic charging curves at various.currents were recorded on a Varian model G-14A-1 strip chart recorder, via a high input impedance (100 NO) voltage follower. This was a standard operational amplifier circuit and was built in the department electrical workshop. The circuit diagram is snown in Figure 26. The potentiostat supplied constant currents of between 1 and 300 mA, the electrode never being made more anodic than 0.8 MITE) to prevent the formation of surface oxides, and never more cathodic than about 5 - 10 mv(RE) to prevent the evolution of hydrogen (see Chapter 6.2.2). A typical charging curve obtained is shown in Figure 27.

Under these conditions the cathodic and anodic charging curves were of identical form. The correction for double layer charging . was made by extrapolating the linear double layer portion of the curve to the potential where the current was reversed. The lengths of the, arrests so determined varied with current. The cathodic curves decreased monotonically with increasing current, whilst the anodic curves usually went through a minimum. At low currents the cathodic value was greater than the anodic one, and the two curves crossed at -2 between 40 and 70 mA (for W'111 A110 mg cm ). Since these measurements 13 on platinized platinum are reckoned to be best at moderate current densities where little or no hysteresis occurs and the system is virtually in equilibrium9b throughout (see Chapter 6.2.2), we chose the point of intersection of the anodic and cathodic curves as giving the best value of arrest length for characterizing the area. The value of S (in cm2) was calculated from the length of the hydrogen -2 arrest (in pc) by means cf the conversion factor of 280 pc cm . This figure is. the one derived empirically98 by.comparing the areas of platinized platinum electrodes obtained by the krypton BET method with the lengths of the hydrogen arrests anodically determined in 0.5M sulphuric acid. Since 1M perchloric acid was used in the present area determinations, this figure is probably more applicable tnan the one -2 98 of 218 pC cm found for 1M hydrochloric acid.

After each area determination the electrode was washed, dried o at 110 C for 30 min, and weighed so that the mass of deposit could • be obtained. 108 Figure 26. High input impedence (100 Mn) voltage follower

4-15V

trim

0/p

Cell loo Recorder Common 0 V

—15V

Philbrik PU8SAY Operational ArnpliPiet-

Figure 27. Typical charging curve for platinized platinum in 1M percnloric acid

1 ----: .---: _.

• —1 t :

....,... W F h t-44 44-44-4- F f r , , 4- r---

4 , ---1 r C t-

-7C

t—

4 I --

a 4 109

Tne cnloroplatinic acid was supplied by 12111. The perchloric acid was BM AristaR, and the lead acetate, hydrochloric acid :aid nitric acid were of AnalaR quality. All water was doubly distilled, the second time from alkaline permanganate solution.

8.2 RESULTS • The results are summarized in Table 14; V is the deposition potential, Y the coulombic efficiency and < i >tile average current density of platinization. Unless stated otherwise, potentials are quoted versus NILE. The potential of the saturated quinhydrone 127 o electrode in 0.144M hydrochloric acid at 25 C was taken as 643 mV 121b (based on f + = 0.78) or 699 mV(RHE), and in 2M hydrochloric H acid as 717 mV (based1211) on f + = 1.01) or 692 mV(RHE). For a H varying current of I flowing for a total time t, the average current density of platinization is given by

< i > = (1/At) j Idt (31) and was evaluated graphically.

During deposition at 394 mV and below, the current tended to fluctuate rather wildly, by up to 10%, particularly in the initial stages of platinization. Below 144 mV the current rose rapidly and then remained approximately constant whereas at 144 mV and above, after an initial steep rise, the current showed a downward trend with time. This became more marked the more anodic the potential.

During platinization at -56 mV the temperature of the solution rose, much gas was evolved (presumably hydrogen) and a fine black suspension was produced in the solution. Hydrogen was evolved at potbntials as anodic as 119 mV, where a little gas was evolved just at the beginning of the plating. With the zero-lead plating solution hydrogen was evolved at 44 mV but not at 144 mV, nor with the 2M HCI solution at 218 mV. However, with a foil electrode (pretreated as for platinization) in D.144M hydrochloric acid alone, hydrogen was not seen until about -66 mV. 110

Table 14

Data for platinized platinum electrodes prepared at various potentials at 25°C

The plating solution for all except the last three was 0.072M chloroplatinic acid plus 1.3 x 10-4M lead acetate in water.

V/mV W /mg /cm2 < i ,/mA Appearance m-2 m ( NHE) cm mg-1 cm-2

-56 11.2 90 1001 49.1 74.9 black, powdery +44 9.8 102 1001 36.2 89.7 black, powdery 5.5 104 573 39.8 91.5 black, powdery 119 9.8 104 1019 35.5 91.3 blck, powdery 144 9.5 127 1196 27.2 81.8 black, velvety 194 10.7 126 1354 24.9 83.6 black, velvety 294 11.7 123 1440 20.0 80.3 black, velvety 394 9.4 91 856 9.0 69.4 dark grey, rough 5.1 92 466 10.0 66.8 medium grey, rough 494 5.0 64 319 2.5 42.1 light grey, rough 594 1.2 100 123 2.4 14.4 light grey, smooth

44a 9.8 93 908 37.5 85.7 medium grey, rough 144a 7.7 129 988 11.8 71.7 bright, smooth 218b 4.8 101 484 16.1 32.7 black, velvety a Plating solution was 0.072M chloroplatinic acid in water. b Plating solution was 0.072M chloroplatinic acid plus 1.3 x 10 4M lead acetate in 211 aqueous hydrochloric acid. 111

At 294 mV and more anodic potentials the solution in the working electrode compartment turned orange-red, and progressively more so with increasing potential. This was due to the accumulation of Pt II species. A red colour developed also with the zero-lead plating solution at 144 mV and the 2M 1-IC1 plating solution at 218 mV. At the more cathodic potentials the plating solution remained a golden yellow.

During a few platinizations, chiefly at the lower potentials, a little deposit (< 4%) fell off the electrode. This was weighed and the value of roughness factor corrected accordingly. The grey deposits formed were somewhat patchy and the light grey ones also distorted the platinum foil substrate a little. •

8.3 DISCUSSION

Figure 28 shows the variation with deposition potential of mass specific area, coulombic efficiency and average current density of 113 platinization. The results of Podlovchenko and Petukhova have also been plotted. To convert their deposition potentials to the NUE scale, the hydrogen ion concentration was taken as 1.5M and the hydrogen ion activity coefficient as unity.

The three solid curves of Figure 28 may conveniently be divided into three deposition potential regions: more anodic than about 300 my, between 300 mV and 140 my, and more cathodic than ea. 140 mV. These regions correspond to different appearances of the deposits. Above 300 mV they were grey, between about 300 and 140 mV black and velvety, while those prepared below approximately 140 mV were black and powdery.

8.3.1 MASS SPECIFIC AREA

The mass specific area has been plotted in Figure 28 rather than the roughness factor since the latter varies almost proportionately with mass degree of platinization W , and this quantity is slightly M different at each potential. The mass specific area, on the other hand, was found to be independent of the mass degree of platinization -2 in the range ca. 5 - 10 mg cm, as can be seen from Table 14 and Figure 28 Tne variation of mass specific area (am), coulombic efficiency (y), and average current density (), with deposition potential (V). This work; 3.5% chloroplatinic acid plus 0.005% lead acetate in water. Circles, 6 ; squares, Y; triangles, . m -2 Open symbols: w1111 = 10.3 mg cm . -2 Half-filled symbols: W = 5.2 mg cm . -2 Filled symbols: W = 1.2 mg cm .

Ref.113; 2% chloroplatinic acid in 1.5M aqueous sulphuric acid.

X, alm ; + Y; *1 . "3

•

• 160 •

E • N

X20

80 ‘dk

.... .666 69s ca 00 0 II

10.01.0.

842

A CC 4-41 0 c.4 4-0 E

116 E ara • 0 • Iri I II Z...1...k,. i f ri -rz.,,17.-...cr...--..^..A...r....rok._ st.vc..-.H.T.Al z: A .1,4.-vwev.i...... 4..w.vc.-. ...i=..s. 0 200 400 600 V/ rnV NH E 114

Figure 28. The results of Podlovchenko and Petukhova are also shown, for comparison.

At 594 mV, W is very much lower than at the other potentials. It will be remembered (Chapter 7.1.1 ) that as Willi decreases, the mass specific area rises slowly at first and then more rapidly, and this may well explain why the corresponding value of am is nigh. Moreover, in this area determination, and also in that of the point at 494 mV, the hydrogen arrest-current curves did not cross but tended to constant values of arrest length. ' The latter were used to calculate the areas.

As the deposition potential becomes more cathodic the mass specific area rises with the rate of deposition, and reaches a plateau level between 300 and 140 mV in Figure 28. The sudden drop in mass specific area at even more cathodic potentials was also observed by 113 Podlovchenko and Petukhova, as can be seen, and was noted more 112 61 qualitatively by Prigent and by Koch (see Chapter 7.1.4). This 61 last worker suggested that a high coverage of hydrogen on the platinum surface inhibits the nucleation process, with a consequent increase in particle size and smaller surface area. Verification of this idea is desirable, and must await a thorough study of the growth processes on platinized platinum electrodes. At potentials more cathodic than about 100 mV the mass specific area continues to decrease, but more slowly, presumably because of further inhibition of platinum nucleation.

The rather large quantitative disagreement (Figure 28) in mass specific area between the present work and that of Podlovchenko and 113 Petukhova, (P and P) may in part be attributed to the composition of the plating solution. Although P and P found that preparation of the deposits from 0.041M chloroplatinic acid solution alone (without their usual 1.5M sulphuric acid) had no appreciable effect on the deposit characteristics, it is not clear whether this includes area. Moreover, 36 P and P did not use lead acetate. Since lead assists nucleation, it will counteract the inhibitory effects of a high hydrogen coverage (at large acid concentration) and is thus likely to give nigher mass specific areas at cathodic potentials.

Another difference between us concerns the technique of area 113 .determination. P and P determined the electrode areas by means 115

of the anodic hydrogen arrests taken in 0.5M sulphuric acid. If -2 they used tne conversion factor of 210 IC cm , which is given in 81 their values of G must be multiplied by 0.75 their references, m to put them on the same basis as the present work. This reduces the two area figures above 150 mV to nearly the same values as ours. However, it also increases the disparity at potentials below 75 mV. The most cathodic point of P and P is probably not trustworthy. For their electrodes prepared at -50 mV(RHE) the anodic hydrogen arrest length showed considerable dependence on the polarising current: when the current density (based on geometric.area) varied from 2.5 to 0.04 mA cm-2, the arrest increased approximately three-fold, -2 113 and the value at 0.5 - 1.0 mA cm was chosen. This behaviour was not manifested by electrodes prepared at deposition potentials above 0 V(RHE). Hence the minimum in their Gin curve is in doubt. The low magnitudes of the values below ca. 75 mV, however, remain to be explained, particularly if the factor of 0.75 is applied to them. Further experiments with plating solutions of varying lead content may clarify this issue.

8.3.2 COULOMBIC EFFICIENCY AND CURRENT DENSITY

The coulombic efficiency in Figure 28 goes through a maximum with increasing potential. To understand this and related behaviour 2 we must correlate it with the. electrode kinetics of the PtC16 reduction process, which involves the following (irreversible) reactions (see Chapter 2.1 and Table 1):

2- - 2- o PtC16 + 2e -' PtCl4 + 2C1 ; E = 0.77 V (1) 2- - - o PtC14 + 2e -4 Pt + 4C1 ; E = 0.75 V (2) o PtC162 - + 4e- -4 Pt + 6a- ; E = 0.76 V (3)

2 The processes that occur during electrolysis of the PtC16 plating solution may be described as follows. At the beginning reaction (1) 2- only is operative because the initial PtC14 concentration is zero. 2- The partial current due to reaction (2), 12 , will then grow as PtC14 ,2- is produced, but increase less than proportionately to the PtC14 •• 27 concentration because the concomitantly formed chloride ions decrease the rate constant of (2). In our plating solutions the chloride 116

concentration rose from alMost zero at the start to more than 0.08M -2 at the end (for the deposition of 10 mg cm ).

At cathodic applied potentials the electrochemical rate constants of both (1) and (2) are large, but the real rate constant of reaction 2 (1) is considerably less because diffusion of PtC16 to the electrode becomes rate-determining. Reaction (2) is not subject in the same 2 way to this limitation as many of the PtC14 ions formed at the electrode surface can be directly reduced in situ. The overall process is then effectively reaction (3). In the anodic region, on the other hand, the electrochemical rate constant of (1), and 2- hence are small, and PtC14 is formed very slowly. This makes 12 very small, since the rate constant of reaction (2) is already low 2 at anodic potentials. Consequently PtC14 , produced slowly, reacts even more slowly, and so builds up in the solution.

This mechanism is consistent with the present results. At deposition potentials above 300 mV the colour of the plating solution II s becomes orange-red as Pt pecies accumulate, and the coulombic efficiency is correspondingly low. As the potential decreases both reactions (1) and (2) speed up, with (1) approaching diffusion control. The coulombic efficiency and current density thus increase, until between 300 and 140 mV the current reaches a limiting value. Here the coulombic efficiency attains a high plateau value (> 80%). At potentials a little cathodic to 140 mV a new reduction process occurs: hydrogen evolution. The sudden increase in current and in coulombic efficiency can be explained by the stirring of the Nernst layer around the electrode by small numbers of gas bubbles. At potentials more cathodic than approximately 100 mV the increasing hydrogen evolution gives rise to a still greater current and causes the coulombic efficiency for platinization to fall. 113 Podlovchenko and Petukhova do not give a current-potential curve and only state values of current density at two deposition potentials (the stars in Figure 28). The absence of stirring accounts for the low values of their current densities. Moreover, smaller current densities would be expected because the plating solutions contained no lead acetate, an additive that snifts25 the current- potential curve to more anodic potentials. The more rapid drop in 117

P and P's coulombic efficiency at cathodic potentials is probably caused by their use of plating solutions 1.511 in sulphuric acid wnich would give rise to greater nydrogen evolution.

It is not easy to compare the present coulombic efficiency figures (Table 14) with the literature data snown in Table 15 because previous workers have not cited potentials of deposition but ratner current densities. It is of course true that currents of reduction processes rise as the potential becomes more cathodic, but there is no simple correlation because, at a given current, the potential of deposition depends on the stirring conditions. The second difficulty is our ignorance about the effect of lead on the electrode kinetics, and virtually all past workers have added lead in various amounts, and sometimes hydrochloric acid, to the plating solution (cf. Table 15). Nevertheless, the coulombic efficiencies lited also pass through a maximum as the current density of plating increases.

8.3.3 DEPOSITION FROM DIFFERENT PLATING SOLUTIONS

No hydrogen was evolved from the plating solution made 2M in hydrochloric acid at 218 mV (ca. 200 mV(RHE)), nor from the usual plating solution without added acid at 144 mV (also 200 mV(RHE)).' A strict comparison is difficult because a high chloride concentration inhibits27 reaction (1) and especially reaction (2), a point confirmed by the fact that the high acid plating solution turned quite red during platinization, and the coulombic efficiency was correspondingly low (Table 14).

Use of the plating solution without added lead acetate affected the appearance of both electrodes (Table 14), while the coulombic efficiency and the average current density of platinization were lower in the case of the electrode prepared at 144 mV. These last two 25 observations are consistent with Bernard's finding that lead acetate caused the current-potential curve to shift anodically by 100 - 200 mV for potentials above about 40 mV. A more surprising result is that the absence of lead produced no significant effect on the mass specific area. Most previous work suggested that a trace of lead enhances the 46 area. Thacker, for example, found the area to increase by a factor of 1.8 when he used a solution of 0.0.51M chloroplatinic acid plus 118

Table 15

Coulombic efficiency data for the platinization of platinum

Chloroplatinic b Lead acetate HC1 CoulOmbic Ref. Current a density acid cone. conc. b conc. efficiency mA cm-2 M x 103 % m x 103 ri ,0

10 c 2.5 51 0.0-0.2 0.0-5.3 0 4o 46 10 3.5 72 0.2 5.3 • 0.1 6o Chap.5.3 3o 3.5 72 0.005 0.13 2 100 Chap.5.3 120 5 103 0.02 0.53 0' 78 62 170. 5 103 0.03 0.79 0 34 105 625 1 21 0.0-0.15 0.0-4.0 0.1 3 54

a These current densities are all given per geometric area of the substrate electrode. b Most workers have given solute concentrations in % w/v. These have been converted to molarities on the basis that chloroplatinic 11 acid is 40 mass % platinum, and that the lead acetate has a stoichiometry of Pb(CH3CO2)23H20 (molecular mass 379.35).

With current reversal every 1.5 minutes.

Cited in Ref.46 although examination of Ref.54 revealed no mention of coulombic efficiency. 119

-2 8 x l0-5m lead acetate, at 10 mA cm for 1 pour with current reversal every 1.5 minutes. However, the work was carried out under galvanostatic conditions at constant coulombic degree of platinization, and is therefore not strictly comparable with the present measurements performed under potentiostatic conditions at constant mass degree of platinization. Although we found the area to be unaffected by the absence of lead, there is evidence that lead enhances the stability of the deposit (see Chapter 7.3).

8.3.4 COULOMBIC SPECIFIC AREA 2 -1 The coulombic specific area (in cm C ) can be calculated from 2 -1 the mass specific area (in cm mg ) and the coulombic efficiency (as a fraction of unity) by the equation:

a 0.5055 G Y (32) c m

Figure 29 shows the variation of coulombic specific area with deposition potential, and the results of Bernard57 are plotted for comparison (see Figure 20). For a given potential of deposition, he found the coulombic specific area to be independent of the coulombic degree of platinization (Figure 15).

Both Bernard's results and those in Table 14 show a maximum at about 220 mV in Figure 29. The quantitative disagreement between the curves is probably due to the fact that the differential capacitance method of measuring electrode area is not usually satisfactory for finely divided platinized platinum, and may give anomalously high 82 results (see Chapter 6.2.2). Moreover, for smootn.platinum, Gilman -2 has criticized capacitances of 20 pF cm as being too low and indicative -2 of surface contamination, values of about 100 pF cm being more typical.

8.3.5 VARIATION OF THE PARAMETERS WITH CURRENT DENSITY OF PLATINIZATION

If mass specific area and coulombic efficiency are plotted against the average current density of platinization, the smooth curves in Figure 30 are obtained. Such curves are, of course, dependent on the stirring rate. It can be seen that the mass specific area is -2 a maximum at a current density of about 27 mA cm . The area data 13 of Slygin and Frumkin for galvanostatically prepared deposits are

120 Figure 29. Tne variation of coulombic specific area (0c) with deposition WiirWA;(11:_id cnloroplatinic acid plus 0.005% lead acetate in. water. -2 w : 10.3 mg cm Open circles: -2 Half-filled circles: w = 5.2 mg cm . Filled circle: w =m1.2 mg cm-2 X Ref.57. to 2.4% cnloroplatinic acid plus 0.01% lead acetate in water.

T .15O

U 100 momomplreem.

0 200 400 600 \i/ mV(N Id _121 Figure 30. The variation of mass specific area (a ) and coulombic efficiency m (Y) with average current density of deposition (ci>).

120

E.") E 80

4 0

8 c)

1 _I 20 'Wm A cm-a This work; 3.5% chloroplatinic acid plus 0.005% lead acetate in water. -2 Circles, a cluares, Y. Half-filled symbols: = 5.2 mg cm ml. -2 m -2 Open symbols: WIni = 10.3 mg cm Filled symbols: W = 1.2 mg cm M - - - - Ref.13: 2% chloroplatinic acid in water at constant currents, - -- -- giving •w- 20.9 mg cm . ___ . . m Crosses. - -

122

Figure 31. The variation of coulombic specific area (0 ) with c average current density of deposition ().

40 0 E 0

20 40 /mA cm/

Plating solution: 3.5% chloroplatinic acid plus 0.005% lead acetate in water.

Open circles: W = 10.3 mg cm-2

Half-filled circles:W1111 = 5.2 mg cm-2

Filled circles: W = 1.2 mg cm -2 123

also shown in Figure 30 for comparison (Figure 18). The disagreement between their work and the present one must be attributed to physical and chemical differences in preparation. 36 Roth and Lasko found the roughness factor of electrodes plated at constant coulombic degree of platinization to be a maximum at -2 approximately 20 mA cm (see Chapter 7.1.3). Since we was constant, it follows from eqn.(20) that AS cc Cre and that ac also exhibited a maximum at this current density. A plot of 0 against the average c current density of platinization for the present work (Figure 31) -2 shows a maximum at approximately 25 mA cm , in fair agreement with the results of Roth and Lasko. It should be pointed out that these workers found a graph of deposit mass against current density to have the same form as the plot of roughness factor against current density. Thus the shape of the latter curve will depend not only on variation of mass specific area with current density of platinization but also on the variation with current of coulombic efficiency.

8.3.6 HYDROGEN EVOLUTION

The evolution of hydrogen during platinization, at potentials more anodic than the reversible hydrogen potential by up to 0.14 V, 22,23,30 has been widely observed. To account for this apparent deviation from thermodynamic behaviour it was suggested30 that hydrogen is formed at very low partial pressures. Such reasoning might explain a small surge of current in a current-voltage curve or a transient in a chronopotentiogram, but it is inadmissible if hydrogen bubbles are actually seen. The final state of the gas must then be 1 atmosphere, and the only feasible explanation consistent with reality and thermo- dynamics must be based on the coupling of chemical reactions, as happens in biological systems.1 •28

One possibility, suggested by Dr D.O. Hayward, involved the following pair of reactions:

H + e H ... Pt (33)

pH + pe- (P/2)H2 (1 atm) (34) which give on summation

(p + 1)11+ + (p 1)e- -3 11 ... Pt + (p/2)H2 (1 atm) (35) 124

For reaction (35) to occur.

LG35 = t.033 + W34 0. (36)

At electrode potentials anodic with respect to RHE, G34 > 0, and therefore LG33 < 0 and kG33 I Z k°341. Reaction (33) is known to occur at potentials from 0.35 to 0 V(MIE) in the charging curve of platinum in an inert electrolyte (viz. the hydrogen arrest used in area determinations). Sub-monolayer deposition at underpotentials has also been reported for silver on 129 platinum (at underpotentials up to 0.25 V) and for copper on 130 platinum (at underpotentials up to 0.48 V). In the present work, hydrogen was seen to be evolved, at least initially, at 119 mV (175 mV(RHE)), continually at 44 mV (100 mV(RHE)), and vigorously at -56 mV(O V(RHE)). Now, with an underpotential 7-1 of, say, 100 mV,

= nF rp = 9.6p kJ. 131 132 Flash desorption experiments and other data for the reaction

1112 (g) + Pt(s) H ... Pt (37) indicate that

6137 = -42 kJ, £S37 -44 J K-1, t1G37 g..1 -29 kJ, and therefore that •

L 33 = 9.6 - 29 kl " 19 kJ.

Thus if ['L35, is to be negative or at least zero, p cannot exceed two. A simple calculation shows that all the hydrogen atoms forming a 2 100 15 -2 monolayer on a 13 cm platinum foil, assuming 1.3 x 10 atoms cm on the platinum surface, would together produce a bubble of hydrogen -4 3 gas at 1 atm only 7 x 10 cm in volume. This would give a bubble approximately 1 mm in diameter. Since more than this was evolved, another set of coupled reactions must be sought.

The two reactions that certainly occur when hydrogen is produced during platinization are:

PtClr + 4e- Pt + Ea- (3) + plH + - (p'/2)H2 (1 atm) (38) which together give 125

+ 2- , . , p111 PtC16 + lP' 4)e -4 kW/ 2)H2 + Pt + 6c1.- (39)

For reaction (39) to occur

6'39 = LG38'5 0. (40)

Thus / 3 < 0.and ILIG3 1 Now

L 3 = - nFE° = -293 kJ,

= 9.6 p' kJ (at an underpotential of 100 mV).

Thus p' can be as large as 30 and still leave bG39 negative. The total amount of platinum deposited in these experiments was approximately 130 mg, or 666 prno1, so thermodynamically permitting the simultaneous formation of 10.0 mmol or 244 cm3 of molecular hydrogen at 1 atmosphere o and 25 C.. This is numerically more than adequate to account for the volume of gas generated during plating at 44 mV. However, two major problems remain. First: why, at fairly anodic potentials, was hydrogen evolved only initially? It cannot be caused by lead covering the growth points in the deposit and so inhibiting hydrogen formation, since a similar phenomenon occurred when plating was carried out without lead acetate at +44 mV. Second: by what mechanism are the reductions 2 of PtC16 and H+ coupled? Reaction (39) is a synthesis of two independent reactions, and the coupling is notional only. One method by which interaction could arise is through a species which involves reagents from both reactions (3) and (38), such as HPtC16 . 133 Although measurements suggest that H2PtC16 is fully dissociated below 0.025M, it is likely that at the concentration of 0.072M used in the present work some HPtC16 ions are present. However, it is far from clear why a species such as HPtC16 - should split up at the electrode surface to form platinum and hydrogen gas rather than platinum and hydrogen ions.

The formation of hydrogen at potentials more anodic than the 102 reversible hydrogen potential has also been reported for a PTFE- bonded platinum black fuel cell electrode in 85% orthophosphoric acid 102 under nitrogen at 1500E. This, too was attributed to the formation of hydrogen at a very low partial pressure, due to the electrode structure permitting very rapid diffusion of hydrogen away from the electrode surface. .It is not clear, however, whether hydrogen gas Was actually seen to be evolved or whether its formation was surmised from a current surge in the current-voltage curve. 126

CHAPTER 9

SURVEY OF RECOMMENDED PLATINIZING PROCEDURES

Both undergraduate laboratory textbooks and certain research monographs recommend procedures for obtaining good platinized platinum electrodes. We shall now examine these in the light of the evidence presented in the previous chapters.

Standard American and British undergraduate texts for practical physical chemistry list a variety of procedures. Daniels, Mathews, 134 Williams, Bender and Alberty recommend a 0.021M (1%) solution of platinum chloride, the use of two dry cells and tile passing of current for several minutes. Shoemaker and Garland135 instruct the student to use a 0.062/4 (3%) solution of platinum chloride plus 5.3 x 10 3m (0.2%) of lead acetate, with two dry cells and a rheostat to give slow gas evolution, the current to be stopped as soon as the electrodes 136 are sooty black. Findlay and Kitchener advocate a solution 0.041M (2%) in platinum chloride and 5.3 x 10-4M (0.02%) in lead acetate for 136a conductance electrodes, and a solution 0.021M.(1%) in platinum 136b chloride and no lead acetate for e.m.f. electrodes, two accumu- lators and a rheostat to give only a moderate stream of gas, reversing the current every half-minute for approximately ten minutes. A very 136a similar preparation to the former is given by James.137 Of these two134,136b 135 recipes do not mention lead acetate, one advocates 136,137 excessive lead acetate, two others support the needless procedure of current reversal, none specify the current density, and 136 only Findlay and Kitchener gives a time of platinization (but again without stating a current density). All the above books incorrectly call the main chemical platinum chloride instead of chloroplatinic acid, so perpetuating a source of confusion found in the older literature (cf. Chapter 1.2).

Probably the most widely used research procedure is that of 138 4o Bates, as recommended by Hills and Ives. This consists of 127

2 passing a current of 100 - 200 mA per cm of substrate for 1 - 3 minutes in a solution of 0.021 - 0.062M (1-3%) chloroplatinic acid 138 plus 2.1 x 10-3M (0.08%) lead acetate. Bates claims that with a "properly prepared" solution the conditions of plating arc not critical. In the light of the evidence available, the present author believes that the lead acetate concentration recommended is too high, and that the range of other conditions given is too broad. The preparation of platinized platinum electrodes adopted should be consistent with their subsequent use.

There are three main uses of platinized platinum electrodes: as hydrogen reference electrodes, as inert surfaces of high area in conductance work, and as catalysts and electrocatalysts. For routine platinization applicable to all three uses, the present writer recom- mends the use of the following conditions: a solution of 0.072M (3.5%) chloroplatinic acid plus 1.3 x 10-4M (0.005%) lead acetate, at a -2 current density of 30 mA cm for up to 10 minutes. A deposition 40,119 time of 5 minutes should be adequate for hydrogen e.m.f. and for conductance7 electrodes, for here a smaller deposit speeds equilibration and reduces adsorption. Good stirring is essential and no gas should be evolved at the platinum cathode. The chlorine evolved at the anode can easily be prevented from interacting with the cathode by employing a salt bridge or an Et-type plating cell. If other considerations dictate the use of a one-compartment plating cell, the evolution of chlorine can be avoided by making the electrolyte 2M in hydrochloric acid and employing a large silver anode, previously lightly cnloridized to prevent oxidation by chloroplatinate ions.

Potentiostatic platinization, a more definitive method of preparation than galvanostatic deposition, is to be preferred whenever the properties of the electrode play an essential part in the research: in electrocatalytic investigations, for example. Suitable electrodes have been obtained47 by plating at 4-50 mV(NHE) from a 0.041M (20) chloroplatinic acid solution, 1M in hydrochloric acid. There was 139 no simultaneous hydrogen evolution. However from the work discussed in Chapter 8 using a plating solution of composition 0.072M chloro- -L platinic acid plus 1.3 x 10 11 lead acetate, maximum area will be obtained by platinizing at +150 mV(NHE), the average current density 128

-2 being approximately 27 mA cm with moderate stirring. These are the conditions recommended by the present writer.

For systems sensitive to trace impurities, such as the hydrogen electrode in a medium of neutral pH, platinization without lead 40 additive has been recommended. In this case a low current density (10 - 20 mA cm-2) is essential for good adherence of the deposit, or else potentiostatic deposition as described above can be used. However, there are two attractive alternatives. One is to add lead acetate as before and, after platinization, to remove the surface lead which is capable of dissolving by soaking the electrode for 24 hours in aerated 1M perchloric acid (Chapter 5). The other is to apply an oscillating signal to the platinum substrate in an inert solution of similar composition to the one to be used. Suitable conditions might be the application of 1 kHz' of amplitude 500 mV centered on 0.95 V for 5 minutes (Chapter 3.4). -Bright and highly active electrodes, but of snort lifetime, can be prepared50 by plating from solutions of chloroplatinous acid. The appropriate pre- and post-treatment of the electrodes has been discussed in Chapters 2.3 and 3.2.3 respectively. 129

PART IV

HETEROGENEOUS CATALYSIS IN SOLUTION 130

CHAPTER 10

SURVEY OF PAST WORK

The vast majority of catalysis work in the past has been concerned either with heterogeneous reactions in the gas phase or homogeneous reactions in the liquid phase. Little attention has been directed at the other two permutations. Yet there are numerous incidental references to heterogeneous catalysis in 140 solution, and in fact this is a growing field of study.

10.1 ELECTRON TRANSFER REACTIONS 141 More than seventy redox reactions have been'examined for catalysis by platinum metal in various physical forms. With only one definite exception, the results fit a mechanism of electron transfer between reductant and oxidant via the platinum metal. The extent of the catalysis should therefore be predictable from purely electrocnemical experiments, involving the reductant - Pt and Pt-oxidant electron transfers, and this has been proved to 142 apply quantitatively to the ferricyanide-iodide reaction. 141 It follows tnat catalysis will be observed if both couples are electrochemically reversible or, wnen,one or both. are irreversible, if the Nernst potentials of the two couples are far apart. This can be shown by a general theory involving the current-voltage curves of the two couples involved and the assumption that these are .additive when both couples are present simultaneously. ,The only unequivocal exception to the predictions of the theory is the rapid catalysis by platinum of the ferricyanide-thiosulphate reaction, in spite of the irreversibility of the S4062-/S2032 system and the small difference of 0.26V in the E° values of the two couples. This 141 may perhaps be attributed to the metal catalysing the reaction by stabilizing some adsorbed intermediate, and not by acting as an electron bridge. Indeed, the ferricyanide-thiosulphate reaction is kinetically very complicated and almost certainly involves radical intermediates. 131

10.2 ISOMERIZATION REACTIONS The heterogeneous catalysis of isomerization reactions of metal 143a complexes is well known. Most work has been concerned with the racemization (optical isomerization) of enantiomorphic complexes of cobalt(III) with ethylenediamine catalysed by charcoal. The bulk 144 of the evidence seems to favour a mechanism whereby a trace of the II corresponding Co complex is produced, and the catalysis is essentially II that of electron transfer froth the configurationally labile Co III species to the inert Co species. The degree of this heterogeneous 145 catalysis is also affected by the presence of added inert electrolytes, particularly on the anion.

Racemization of complexes of some other metals, e.g. CrIII 146 III 147 and , are not catalysed by charcoal. Since the ammine complexes of these two metal ions are not readily reduced, this 143a II observation indirectly supports the view that Co is required III 143a in the charcoal catalysis of Co reactions. Other workers have tried out solids such as Raney nickel, platinum, palladium, rhodium, silica gel, alumina and zinc oxide. The platinum group metals appear to be the only other effective catalysts besides charcoal.

The effect of polyelectrolytes can be considered to lie between the homogeneous and heterogeneous categories. The rate of racemization of tris(o-phenanthroline)iron(II) perchlorate was decreased 48 by the presence of sodium polystyrenesulphonate, abruptly at a low concentration of the polyelectrolyte, and as this concentration was increased it snowed an almost total lack of dependence. It was proposed that the racemization occurred by ligand exchange and that sorption on the polyelectrolyte therefore impeded this intermolecular reaction.

10.3 SUBSTITUTION REACTIONS The heterogeneous catalysis of the reaction: + + C2H5I + Ag + 112 0 C2 HSOH + AgI + H (41) , 9,150in has been studlea14 detail. It is strongly catalysed by gilver halides, other silver salts, and charcoal, only weakly catalysed by finely divided platinum, palladium, silver, silicon and silica, and not catalysed at all by powdered glass and barium sulphate. The kinetics fit a Langmuir-Hinshelwogd mechanism. The heterogeneous catalysis of the hydrolysis of haloalkanes and 151 2-haloalkanoic acids has been reviewed. 132

Inorganic ructions have also received some attention. The heterogeneous catalysis by a number of solids _of the aquation reaction: 2+ Co(NH3)5X + H2 O -, Co(NH3 )5(H20) 3+ + X- rLX = CI or Br) (42)

in 0.01M perchloric acid has been examined by Archer and Spiro and the heterogeneous catalysis of other inorganic substitution 152 2 reactions was reviewed. The catalytic effectiveness per cm of 152 the solids were: for X = Cl, Pt >> AgC1 >Hg2 C12 > AgBr > AgI >HgS; and for X = Br, Pd ^• Pt > AgCl > Au > AgBr > AgI > HgS >rigiBr2. PbS, BaSO4 , glass, and stainless steel had no effect at all. Catalysis by the silver halides was subject to a considerable photochemical effect, and silver, mercury and charcoal caused rapid reduction to 155 give cobalt(II). The reduction by silver was investigated quantitatively by means of silver rotating discs, and found to be transport-controlled for both the chloro- and bromo-complexes. The catalyses for X = Br 153 by mercury(II) sulphide and silver(I) bromide powders were examined in more detail and the kinetic data fitted best a model based on a Langmuir-type adsorption preceding the chemical reaction. Similarly, the catalysis for X = Br by a large platinum rotating disc was 154 investigated and the Langmuir model was again found best. In 154 this last paper the electrical potential of the platinum was monitored and found to have ,a definite influence on the rate of reaction. Clearly a platinized platinum surface would have an even more profound catalytic effect. This prompted the present study and the work will be discussed more fully later (Chapter 12).

Finally, a great deal of work has been done on quasi-heterogeneous systems, namely catalysis by polyelectrolytes and micelles. The field has been adequately reviewed for micellar catalysis of organic 156,157 reactions and for catalysis and inhibition of both organic 158 and inorganic reactions in synthetic polymer and micellar solutions. The acceleration or inhibition of chemical reactions in micellar and 156-158 in polymer solutions arises from the partial distribution of the substrate in the micellar/polymer phase as a result of electrostatic and hydrophobic interactions between the substrate and the 157 surfactant. Using simple electrostatic considerations one expects, for example, cationic micelles to enhance the rate of reaction of nucleophilic anions with uncharged substrates (e.g. the saponification. 133

of p-nitropftenylnexanoaie), anionic micelles to retard it, and non-ionic micelles to nave little or no effect. The hydrophobic interaction means that long-chain reactants are incorporated into the micelles to a greater extent than are short-chain reactants, but this effect is much less with polyelectrolytes. 158 The hydrophobic interactions are not present, of course, with inorganic 158 reactions (e.g. the mercury(II) induced aquation of cnloropentam- minecobalt(III) perchlorate catalysed by polyanions). One motivation 156-158 for studying these quasi-heterogeneous catalysts is the analogy with the much more complex enzyme catalyses. 134

CHAPTER 11

THEORY OF THE ROTATING DISC AND ITS APPLICATION TO CATALYSIS

The overall process occurring in any heterogeneous reaction comprises159 five basic steps:

(a) Transport of solute species to the interface. (b) Adsorption at the surface. (c) Reaction at the surface. (d) Desorption of the products. (e) Transport of the products away from the interface.

Of these, steps (b), (c) and (d) are chemical processes, and steps (a) and (e) are transport processes. In principle, any one of the steps could control the overall reaction. With heterogeneous reactions in the liquid phase, transport is obviously more likely to be significant than in the gaseous phase. 159 Transport control in heterogeneous reactions has been discussed for a wide variety of reaction types. The original treatment in this field was due to Nernst, according to wnom chemical processes at an interface are usually much faster than one or other of the transport processes,kso that, unless there is a slow process occurring within the bulk of one of the phases, the observed rate is transport controlled. Nernst assumed159 that for a well stirred solution the bulk concentration of reactant is uniform, but that adjacent to the solid exists a static (Nernst) layer through which diffusion occurs. Thus for a solid of geometric surface area A, immersed in a Volume V of solution, of reactant concentration c, Fick's first law states:

-dc/dt = (DA/V)ac/y (43) where D is the diffusion coefficient of the solute and (c/ay) the concentration gradient normal to the surface. Nernst further assumed tnis concentration gradient to be linear, so that 135

-dc/dt = DA(c' c")/V8 (410

where c' is the bulk concentration, c" that immediately adjacent to the surface, and 5 the tnickness of the Nernst layer. For very fast surface reactions c" tends to zero, and:

-dc/dt = (DA/Vo)c (45)

giving a first order rate constant of DA/Vb.

However, the assumptions underlying this treatment are now 159,160 regarded as unsatisfactory, and when the equation is applied to liquids in motion it should be considered to be only empirical. That the liquid retains some mobility down to very small distances 160 from the surface, much less than 8, does not imply the absence of a diffusion step in the process.

Although for most flow systems the mathematical analysis of fluid flow is difficult, if not impossible, for certain cases a quantitative treatment is possible. The most successful is for a rotating disc, and rotating disc electrodes have therefore been widely used in electrochemistry. However, the hydrodynamics are not restricted to electrochemical systems and, indeed, the principles 142,154,155,161 have been applied - to a wide variety of heterogeneous reactions in the liquid phase. - The rotating disc was therefore selected for the present study, and the hydrodynamical Principles are outlined bselow.

11.1 MASS TRANSPORT IN THE ROTATING DISC SYSTEM 16o-162 The rotating disc system (RDS) involves convective diffusion of the solute species in the liquid solvent, i.e. a combination of molecular diffusion (resulting from chemical potential gradients) and transportation due to entrainment in the moving liquid. Thus, to examine the motion of the solute particles, it is necessary to know the functional dependende of the fluid flow on the spacial co-ordinates, and the hydrodynamics must first be solved.

11.1.1 HYDRODYNAMICS The RDS comprises 160-162 an infinite horizontal lamina rotating with a constant angular velocity, W, in a liquid of infinite extent, 136

as depicted in Figure 32. - The liquid is treated as a continuum, and as incompressible. .The hydrodynamic equations to be solved are the Navier-Stokes Equation:

d71; 1 (46) 71-t - P 4 V C2 -;

and the Continuity Equation:

= 0 (47) where v is the velocity of a volume element (vector), p the density (scalar), p the pressure (scalar), and V the kinematic viscosity (scalar). Natural convection was ignored, and the assumption of incompressibility has already been incorporated in eqns (46) and (47). These equations are put into cylindrical polar co-ordinates and im ,161 s plified16° by means of the following constraints:

(a) only the non-turbulent, stationary state is-considered when aV/at = -6;

(b) because of axial symmetry all derivatives w.r.t. 9 vanish;

(c) v is also assumed to be independent of r;

(d) the pressure, p, is a function of y only.

The boundary conditions necessary for the solution are illustrated by Figure 33, and are:

at y = 0, vr = 0, ve= Wr, vy = 0;

at y =cc, vr = 0, ve 0, vy = -u;

where u is a constant velocity to be found. The resulting system 16o'161 of equations may be.solved by casting them into non-dimensional form, and assuming formal power expansions for the solution. The picture emerging is one whereby, far from the disc, the liquid moves only towards it. In a thin layer adjacent to the disc surface, the liquid acquires a rotating motion, until eventually the radial velocity goes through a maximum and then drops to zero at the disc surface, and the angular velocity of the rotating disc is attained. It is found that v = -0.8u and v0 = 0.05Wr at a distance 60 from the disc 160 Y surface, where 60 , the thickness of the hydrodynamic boundary layer, is given by: 1 so = 3.6(\iw)2 (48) 1.37

Figure 32. Cylindrical Polar co-ordinates for the rotating disc system. 138

Effectively, within this boundary layer the radial and tangential velocity components are not zero, while beyond it only axial motion exists (Figure 33).

11.1.2 SOLUTE TRANSPORT 16o 161 The transport of the solute can now be examined. Consider the disc to be a sink where the reactive species are destroyed. In the case of ionic solutes it is supposed that a great excess of inert supporting electrolyte is present so that electrical migration can be ignored. Adding the diffusional and convective contributions of the total flux gives the Equation of Convective Diffusion. This is put into cylindrical polar co-ordinates and simplified by means of the previous constraints. In addition, attention is confined to the steady state where the concentration distribution is independent of time (ac/at = 0). As for v, c cannot depend explicitly on 9, and it 16o is also assumed to be independent of r (i.e. edge effects are ignored ). The ultimate version of the Equation of Convection Diffusion reduces to one-dimensional transport normal to the disc:

dc tic v — = D (49) Y dy 2 dy where D is assumed independent of c. The boundary conditions are:

at y = 0, c = c", v = 0; 1 at y =Cc, c = cf, v = -D.886 (wv) Y 2- After transforming the variables, the double integration can be rm perfo ed160-162 to give c as a function of y, which is shown in Figure 34. At distances from the disc surface closer than 6, ci approaches c" almost linearly, so 6 is the thickness of the diffusion boundary layer and is given by the Levich Equation:

b=1.61213 w1 1 (50)

with all parameters in c.g.s. units. Comparison of eqns (48) and (50) gives an idea of the relative magnitudes of 60 and 6:

= 0.45 (D/0 80 (51) At a typical value of (DA) of about 10-3, so is 22 times as big as 6, and at 400 r.p.m. (w = 41.9 rad s-1), 6 is approximately 0.0025 cm. 139

Figure 33. Fluid flow in the rotating disc system.

Figure 34. Concentration profile normal to the rotating disc.

y/6 140

1b3 Gregory and Riddiford integrated eqn. (49) graphically and found that for values of D/v from 0 to 0.004, the semi-empirical expression:

8 = 1.613I35- w-1 [1 + 0.354 (131/v)0.36] (52)

gave tnese results to within 1%. This equation has been thoroughly 161 tested, and found to describe experiment better than eqn. (50), • 161 164 although tne disparity is typically only about 3%. Newman has 2 integrated eqn. (49) analytically for values of D/V less than 10 , and derived: = 1.6124 vl mil + 0.298 (D/v)3 + 0.145 (D/v)i] (53)

neglecting nigher order terms in D/v. Tnis agrees with values obtained by numerical integration to within 0.1%. Since v = 0 at y = 0, Fick's first law (eqn. (43)) can be used to evaluate the flux, 160,161 j, to the disc surface. The functional dependence of c upon y gives dc/dy as (c' - c")/8 at y = 0, so:

j = D(c' - c")/8 (54)

Comparison of eqn. (54) with (44) shows that (5 is also the thickness of the Nernst diffusion layer.

11.1.3 FURTHER ANALYSIS The disc Is a uniformly accessible surface because S is independent 161 of r. This arises from• the assumption that c is independent of r, and is contravened in practice%at the disc edge, to which the foregoing analysis does not apply. Radial diffusion in the system comprising a disc electrode of finite size, embedded in an infinite insulating 165 plane has been treated mathematically. Electrical migration was ignored, i.e. the solution was of a neutral species or with an excess of supporting electrolyte. The region where radial diffusion is important is very small (of the order of (6/r)) and located at the electrode-annulus junction, where at the limiting current, back diffusion from the annulus zone to the edge of the electrode results in an increased current density. This region is situated well within the diffusion layer. With a finite rotating disc and no annulus the zone 16(3 where c is not independent of r extends over a distance of about 80.

In addition, a non-uniform current distribution over the disc 166 surface can arise from electrical resistance effects. Newman has 141

examined theoretically the RDE described above, for solutions of a single salt and ionic solutions with excess supporting electrolyte, giving a moderately fast electrode reaction. At currents smaller than the diffusion limiting value the current density varies with r, from a lower value at the centre to a higher value at. the edge, because tne edge of the disc is more accessible than the centre as a result of the ohmic potential drop in the solution outside the diffusion layer. 167,168. This has been experimentally verified for the electrodeposition of copper on a copper rotating disc electrode (RDE). As the size of the electrode expands or the electrode reaction becomes very reversible, this non-uniformity of current density should be enhanced, but as the amount of supporting electrolyte is increased the non-uniformity should 168 subside. This last prediction too has been verified for the electrodeposition of copper on a copper RDE. Moreover, a fifty-fold 168 excess of supporting electrolyte was found to.be sufficient to produce a unifonnly accessible surface provided the current exceeded 20% of the limiting value. 169 Essentially, when the contribution of ohmic resistance in the bulk of the solution to the total resistance is large, the current density distribution will be non-uniform, but when the contribution of charge transfer or concentration polarization is large, then there will be a uniform current density distribution.

11.1.4 PRACTICAL FACTORS . Obviously, the theory outlined above is only an approximation to a practical system, where the disc is perforce of finite size, etc. 161 Operationally, a disc is effectively infinite if its radius is very much greater than the thickness of the hydrodynamic boundary layer, 60. Thus at low rotation speeds the disc must not be too small. Nor can very high rotation rates be employed, for the flow becomes turbulent when tne dimensionless Reynolds number, Re, defined as:

Re = r2 (.11/y (55) 161 is greater than ca 2 x 105. To maintain laminar flow this value should not be exceeded at the edge of the disc. Theoretical design factors and experimental performances of a variety of rotating discs 161 nave been assessed, and a trumpet shape was recommended so that flow 142

patterns above and below the disc interfere as little as possible. 161 The importance of the vessel size has also been discussed. With a large trumpet shaped disc (the type used in the work to be described in Chapter 12) of 7.6 cm diameter the containing vessel should be170 170 at least 17 cm in diameter. For reproducible results there should also be at least a 1 cm clearance from the bottom of, the vessel, but the depth of immersion beneath the surface of the liquid has no 161 significant effect. Similarly, the disc should be well centered on the vertical drive shaft to minimize lateral eccentricity, and the drive shaft must be rigidly held to avoid precession. 161 For en.q (54) to hold any rugosity of the disc surface must be small in comparison with the thickness of the hydrodynamic boundary 171 layer. Fleischer and Schuberth used the reduction of ferricyanide to ferrocyanide •to study the effect of rugosity of a copper RDE on the diffusion limiting current density, and thus characterised the onset of turbulence. They defined a rugosity factor as the average depth of protrudence across the surface, and the critical value giving turbulent flow, Ru(cm), was found to be given approximately by:

Ru = 0.245 (1.14)1 (56),

For a dilute aqueous solution with the RDE rotating at 400 r.p.m. -1% (41.9 rad s ). the critical rugosity factor would be about 0.0036 cm, i.e. the same order as the thickness of the diffusion boundary layer. As stated in Chapter 10.31'the present proposal was to use a large platinized platinum RDE. Very good agreement has been obtained172 between smooth and platinized platinum RDEs for the i/4 plots of the anodic ionization of hydrogen in various electrolytes at values of W up to 2622 r.p.m. (274.6 rad s-1). At this high speed of rotation the current on the platinized platinum disc was significantly higher than that on the smooth one, and corresponds to a critical rugosity factor of 0.0014 cm.

The positioning of a Luggin capillary in electrochemical measure- 161 ments with a RDE has been discussed, and it now seems clear173 that the tip of tne capillary should be close to the actual electrode surface. 114-3

'11.2 KINETIC ANALYSIS

The system to be considered here is the heterogeneous catalysis of a reaction in solution. To follow its progress, small samples of solution are periodically removed for analysis, but the amount of catalyst is considered fixed (as in the RDS); thus the amount of catalyst relative to the volume of solution increases with time. The reaction occurs homogeneously as well as on the surface, so this must be taken into account in interpreting the analytical results. Since it is the catalysed reaction which is of interest, the heterogeneous rate will be assumed to be controlled by the rate of the surface reaction on the catalyst. The case of intermediate kinetics, where the surface rate is fast enough to be comparable with the diffusion limited rate, will receive some.attention in a later section.

11.2.1 SURFACE CONTROL For a first order surface reaction:

-dcs/dt = k s cs (57) where c is the surface concentration of the reactant (mol cm ), k s s the first order surface rate constant, and the product is assumed to be rapidly desorbed. The rate of change of concentration will be due to both heterogeneous and homogeneous reaction, and if the latter is also first order:

-dc/dt = S k + k c (58) - s cs b where c is now the bulk concentration, S the real surface area, V the volume of solution, and k the bulk first order rate constant. b To express this rate in terms of bulk concentration only, an adsorption isotherm must be assumed. Two of these will be used. The reaction must then be slow enough for adsorptive equilibrium to exist.

(a) Henry's Law Adsorption c = ac (59) where X. is the adsorption coefficient; this is only valid at low coverage. Substituting eqn. (59) into (58) gives:

-dc/dt = [(SXks/V) + kb ]c (60)

The volume is only constant between successive measurements. Thus the 'expression can only be integrated over such a period from t to i-1 t. say, giving: 1

114

-1n(ci/ci_ ) = SAk )/vi_ + kb(ti - ti_1) (61) 1 s(ti ti_1 1 However, at any stage i = n (say), the preceding terms can be summed:174

ln(c /co) + kb(tn - to) = - SXks (t. - t. -1 )/ (62) n 1 =1 1 Vi-1

(b) Langmuir Adsorption cs /cm = Kc/(1 Kc) (63)

where cm is the surface concentration at monolayer coverage, and K the adsorption coefficient; at low Kc this reduces to eqn.(59). Substituting eqn. (63) into (58) gives:

-dc/dt = [ScmKkg/(1 Kc)V kia]c (64) Putting, (65) B = Scm Kk s andintegratingbetweentandt.,by means of partial fractions, 1 leads to:

V.1-1 Bi-le V. (14-Kc.1-1 ) 1 V.1-1 l bVi_i :)- t- ti-1 (66) Clcb B+k- b V. 1-1 In ILB+1:013 .13.:1(1+Kc.)..1 B+k ci

If the first term of this expression is ignored, the equation is then merely a re-arranged form of eqn. (61). It is therefore desirable to decompose the first term into a more tractable form. Using the expansion in(1 + y) x for x << 1 gives: -Bfl + k V. (1+ Kc. )/B1 kb-1Vi K 14 b 1-1 1-1 ] (c. - c.) (67) fl V. 1-1 10 (1+ Kc.)/B1 . 1

for kbV(1+Kc) << B; this is equivalent to kbVc << ks Scs which means that the total amount reacting by the homogeneous path must be much less than that by the heterogeneous path. Substituting this approximation into eqn. (66) and re-arranging gives:

1-1-\ ( 68) B+k V. ic(c. b 1-1 L 1 Cross-multiplying this and re-arranging produces:

) = -B(t.-t. (69) ln(c./c.1-1 )+ kb13.(t.-t.-1 1-11 )/V. -1 + K(c.1 -1 c.)1

Summing as before and dividing throughout by (co o n t.-t. ln(cn/c0) + kb(tn-to) . B ( 1 1-1) + - K (70) o n V.1-1 o n 1=1 145

When K is small this reduces to eqn. (62).

The ultimate aim of drawing plots according to eqns. (62) and (70) is to evaluate the rate constant k from the slopes and intercepts. s To do this, certain other parameters (S, X, cm) must be evaluated separately. The real surface area, S, can be obtained by a large number of techniques175 (e.g. Chapter 6). But the others, related to the equilibrium isotherm, present a problem. There are two possible ways of estimating them. Firstly, the values pertaining to a similar yet stable species to the reactant could be used by determining its adsorption isotherm experimentally. Secondly, extrapolating plots 1 of Line + - to n kb(tn ) I against the sigma function to zero time should provide a value of c which is less than the actual c , because some o o of the reactant will have been adsorbed. The difference measures how much is adsorbed, and enables an adsorption isotherm to be constructed, so yielding the required parameters. For the second method to be applicable, the area must be large enough, relative to the volume of. solution, to result in a measurable amount of adsorption.

If the size of the samples withdrawn for analysis is constants a, say, the volume variation is given by:

= Vo -(i-1)LV (71) where V is the starting volume. o 153 176a (c) The Archer-Spiro Model .These workers approached. the problem in a more phenomenological way. The rate of reaction may be expressed in two ways:

-dc/dt.= (kh + kb)c (72)

-dc/dt = Sk c /V + k c (58) s s b where k h is an empirically observed first order rate constant for the heterogeneous contribution. Equating these expressions, substituting for c s from the Langmuir isotherm eqn. (63), and re-arranging gives:

1/kh = (V/Skscm)c + (V/SkscmK) (73) 153176a Operationally, the mean value of c for each run was used, and V was assumed to be constant. The latter was effectively true in those cases in which the catalyst was in powder form and some was then removed in each sample. Equation (73) was found to describe experiment quite well. The difference of this approach is that each run is

146

treated empirically as first order and the variation of kh with c is analysed to give k ; the previous treatments (a) and (b) analyse s the nature of the measured rate constants for each individual run. In eqn. (73) the same number of parameters need to be evaluated as in eqn. (70). 176a A relationship has been derived between c and t for an individual run, based on the treatment above. Putting,

V/Sc k = h m s (74) and substituting eqn. (73) into (72) gives:

-dc/dt = [kb + 1/(ho + b/K)]c (75) This can be re-arranged and integrated, and yields:176a

kbh(1 + Kc) In c + ln k Ic.bh K + bh(1 + (76) Kco

To be able to plot a graph of eqn. (76) requires values of the parameters 176a K and h, and hence S, cm and ks. Its use appeared for kinetic plots over a long time, where a normal first order plot became unsatisfactory. However, it would involve much work to evaluate the kinetic parameters solely by eqn. (76), e.g. by computer curve fitting.

11.2.2 INTERMEDIATE KINETICS This is concerned with the RDS where the rate of chemical reaction at the surface is comparable to the rate of diffusion. The same general considerations as previously (Chapter 11.2.1) pertain and eqns. (57) and (58) still apply. However now, in the steady state, the part played by mass transport of the reactant in the rate of the heterogeneous reaction will be comparable to that of the surface reaction itself. The rate of the surface reaction is always equal to the rate at which reactant arrives at the surface, viz. the flux j, but the surface reaction no longer controls the flux and both rates must be taken into account. As regards mass transport to the disc, the 160 geometric area, A, is the relevant quantity, so equating the flux and reaction rate: Aj = Skscs (77)

Combining eqns. (50, (58) and (77) gives: 147

-dc/dt = AD(c c")/V6 + kbc (78) remembering that c is now the measured bulk concentration which was written c' in eqn. (54). To integrate this expression, the bulk concentration adjacent to the surface, c", must be known in terms of the other parameters. Combining eqns. (54) and (77), and re-arranging:

D(c - c")/6 = Acscs (79) where RS = s/A (15) Further development requires c in terms of c". If this is to be expressed in the form of an adsorption isotherm, cs must again be regarded as an equilibrium surface concentration; i.e. even with fast diffusion and fast reaction, adsorption is assumed to be even faster.

(a) Henry's Law Adsorption C = (59)

Here, c" is the appropriate bulk concentration as far as adsorption is concerned. Substituting eqn. (59) into (79) gives:

c" = DC/(W6 + D) (80) where B' = fOks (83.)

At large 6 (slow stirring) B16 > D and c" as (DVBIOc, so the reaction is almost diffusion controlled. At small 6 (fast stirring) D >> and c" gsc, and the reaction approaches surface control. Putting eqn. (80) into (78):

-dc/dt = UADBIAB1 6 + D)V} + k Jc (82)

As this is of the same form as eqn. (60), a similar integration procedure leads to: n ln(cn/c0) + kb(tn t ) =(-ADB1/016 + Dg) (t. - t. V.-1 (83) o i.1 1-1 )/1

For a fast reaction (diffusion control) B16 > D and the effective rate constant is AP/61 When the rate is slow (surface control) D > B1 6 and the rate constant becomes SXk as in eqn. (62). With s increasing stirring, therefore, 8 progressively decreases until the reaction proceeds smoothly from diffusion control to surface control. 148

c /c = Kc"/(1 + Ken) (b) Langmuir Adsorption s m (63) Substituting eqn. (63) into (.79) gives:

c"2 + [(Bub/DK) + (1/K) c]c" - c/K = 0 (84) where B" = Ace m ks (85) Equation (84) can be put into a simpler and dimensionless form by the transformations:

g = c"/c (86)

a = 1/Kc (87) b = (D + Bub)/D (88) yielding:

e (ab - 1) - a = 0 (89) Hence.:

g = [1 - + j (ab - 1)2 + 4a111/2 (90) The square root.term can be conveniently re-arranged to ab[l + (4/b + 1/ab 2)/abf/ -rd and expanded to ab[l + (4/b + 1/ab 2)/2ab]. since (1 + x)2 ^,1 + x/2 for x << 1. Thus eqn. (90) can be re-written:

g [1 - ab + (ab + + 1/2 ab 1) ]/2 (91) if (1 + 1/4a)/b << (1 + ab/2)/2 (92) From eqn. (92) it can be seen that for (91) to hold, the values of a and/or of b must be sufficiently large. In physical terms, it means that the reactant must be present at low concentration, or else the surface rate must be large enough for the reaction to be almost diffusion controlled.

Two values of g are produced by eqn. (91):

'(1 1/4a)/b (93)

1 ab -(1 + 1/4a)/b (94)

Since a and b are large, and since g has to be positive, g_ will be a highly unlikely solution. From eqns. (86) -_(88) and (93):

c" = (1 + Kc/4)1DC/(kb + D) (95) Putting eqn. (95) into (75):

-dc/dt = pc - qc2 (96) 149

where:

P Pt 4. kio (97)

p'= ArOks + D)V (98)

q = AD2K/48(ko + D)V (99) Integrating eqn. (95) from time ti_i to ti, V only being constant during such a time period:

ln(ci/ci_i) - ln[(p - qci )/(p - qci-1 )] = p(t. - t..- 1 ) (100) Further development requires simplification of the second term on the left hand side. This can be done by putting it into the form ln[{1 + (kb - qci)/p9}/{1 + (kb - qci...1)41}]

and expanding, using ln(1 + .N) i x for x << 1; i.e. 11% - qpi/p1 << 1. If qc > kb, this can be related back to eqn. (96), where because -dc/dt is positive p - qc > 0. Then eqn. (100) becomes:

In(ci/ci_i) - q(ci_i - ci)/p9 = - p(ti (101) - ti-1) Substituting back for p,p' and q from eqns. (97) - (99), and summing from i = 1 to i = n, gives the final equation:

ln(c/c ) + k (t ) Ate" n -t. o b n to iv .1-1) 4. i+B D:C, (102) c c (B"8+D)(c -c ) 0 o n (t 1-1 i=1

This is of the same form as eqn. (70). When 6, is large (slow stirring) the rate constant is ADA and diffusion control obtains. For small 6. (fast stirring) the rate constant becomes Agyand surface control results, although it must be remembered that, in this case, to keep a large, c. must be very small. The latter rate constant is the same as in eqn* (70) when But is substituted from eqn. (85) - as required. 1'0

CHAPTER 12 •

THE HETEROGENEOUS CATALYSIS OF THE AQUATION OF BROMOPENTAMMINE- COBALT(III) BROMIDE BY PLATINIZED PLATINUM

The equation reaction (42) (at pH below 4-5):

Co(NH3 )5X2+ + H2O Co(NH3)5(H20)3+ + X [X = Cl or Br] (42) appears to be a simple substitution reaction which, as mentioned in Chapter 10.3, is heterogeneously catalysed by platinum. For reasons discussed below, a study of the catalytic effect of platinized platinum promised to throw much more light on the process and the research involved is described in this Chapter.

The homogeneous kinetics of reaction (42) are first order and 143b,177 independent of pH in the range 1-4. Opinion favours a mechanism involving solvent-assisted dissociation of the halide anion, 178,179 and recent measurements of the volumes ofactivation support this. Both the reactant and product complexes are octahedral. The equilibrium position of reaction (42) lies well to the right, and in dilute solution (< 10 2M) the forward reaction proceeds virtually to completion (Section 12.2.2).

As stated in Chapter 10.3, the heterogeneous catalysis by a number of solids of reaction (42) in 0.01M perchloric acid has been examined 152-154 by Archer and Spiro. With X = Br, the catalysis by a large smooth platinum rotating disc was studied154 in detail. The following salient points, concerning the catalysed rate of reaction, emerged:

(a) the reaction was surface controlled; (b) the rate depended on the electrochemical pre-treatment of the platinum catalyst; (c) the potential of the platinum was a major factor in determining the rate of reaction - the lower the potential, the faster was the reaction; (d) with the addition of Br and Co(NH3 )5 (1120)34 ions the catalytic rate increased and the.potential of the catalyst was lowered; 151

(e) the kinetics conformed well to a modbl based on Langmuir adsorption prior to reaction (see Chapter 11.2). , Several of the reaction mixtures were analysea176b for cobalt(II) (2-n)+ by spectrophotometry of the chloride complexes CoC1 n = 1-4. n In no case was any found. The dependence of the rate of reaction on the electrical potential of the platinum catalyst is very interesting, and indicates tne possibility of overt control of catalytic activity during the course of reaction.

12.1 AIMS OF THE PRESENT WORK '54 It was pointed out that the increase in catalytic activity caused by lowering the disc potential may be due partly to an increase in the adsorption of the reactant (an increase in K), and partly to a change in the free energy of activation (which affects ks). It would be very desirable to examine these two aspects separately. This is made difficult by the fact that to observe the catalytic effects at all, it was necessary154 to work at very low concentrations ((1 - 15) x 10-5M for an area of 25.16 cm2) otherwise the homogeneous contribution swamped the heterogeneous one.

In principle, the use of platinized platinum should permit this problem to be solved. The greatly increased real surface area should enable the reaction to be followed at correspondingly higher concentrations. This leads to two advantages. Firstly, the presence of cobalt(II) is more easily detected and estimated. Secondly, the amount of reactant adsorbed should be measurable by extrapolating back to zero time (see Chapter 11.2), yielding the adsorption isotherm. Furthermore, by examining the effect of potential on the rate and adsorption, the contributions of changes ink and K should be separable. s The enhanced catalytic effect of a very much larger real surface area at constant geometric area is subject to the accessibility of the former. Since to obtain a large roughness factor necessitates an intricate pore structure, the reactant may have difficulty in approaching much of the increased surface. A measure of this difficulty is the time needed to reach adsorptive equilibrium. A great deal of workao on adsorption has been carried out using radioactive tracers, and the time for adsorptive equilibrium to be attained on platinized platinum for solutions of, typically, 10-3M cadmium bromide, was several hours. 152

The surface state of the platinum is also important: the time taken 180 to reach adsorptive equilibrium on pre-oxidized smooth platinum, for 10 -14 sodium bromide, increased from two minutes after several seconds of atmospheric oxidation, to a few hours after anodic oxidation in 0.5M sulphuric acid. On the other hand, work on the electron exchange catalysis by platinized platinum of the couples thallium(I)- 114 thallium(III), and the tris(ethylenediamine) and the ethylenediamin6- 181 tetraacetic acid complexes of cobalt(II)-cobalt(III) indicated adsorptive equilibrium to be attained in less than 15 s for the former and 3 - 5 minutes for the latter. No generalizations can therefore be made from the literature evidence.

12.2 PRELIMINARY EXPERIMENTS

12.2.1 PREPARATION AND ANALYSIS OF BROMOPENTAMMINECOBALT(III) BROMIDE 182 The compound was prepared by oxidation of cobalt(II) in the presence of ammonia and bromide. To a solution of cobalt(II) bromide (46 g), ammonium bromide (50 g) and concentrated ammonia solution (250 ml) in water (65 ml), was added 30% hydrogen peroxide solution (40 ml): 2+-• lu n + 2112 ../2 51\.11-13 + H2 0 Co(NH3)5(H20)3+ + OH (103)

The excess hydrogen peroxide rapidly decomposed in the alkaline solution, and excess ammonia was removed by bubbling nitrogen through the solution for three hours Concentrated hydrobromic acid was then added until a precipitate of aquopentamminecobalt(III) bromide persisted, a further 50 ml was added, and the evaporating dish was heated on a steam bath for two hours. The crude brOmopentamminecobalt(III) bromide was filtered off and recrystallised.

Crude product (25 g) was added to a hot (90°C) solution of concentrated ammonia solution (25 ml) in water (500 ml). This was subsequently filtered and hydrobromic acid (40 ml) added to the filtrate, which was heated on a steam bath for two hours. The deep purple crystals of pure product were filtered off, washed with water and alcohol, and dried for two hours at 110°C. The complex was analysed for both bromide and cobalt.

For the bromide determination, the complex (0.1 g) was dissolved in 0001M potassium hydroxide solution (100 ml) and left for 30 minutes. 153

This solution was just acidified with nitric acid and titrated with 0.1M silver nitrate solution. The indicators dichlorofluorescein and phenosafranine did not give distinct end-points, so a differential potentiometricl83 technique was used, employing a silver wire indicator electrode. Just before the end point was reached the solution was allowed to stand for 20 minutes in the presence of the silver bromide precipitate, which would catalyse the hydrolysis of any remaining starting material. This gave the amount of'bromide as (62.08 0.21)%1 compared to the stoichiometric value of 62.46%.

For the cobalt determination, the complex (0.1 g) was dissolved in 0.05M sodium hydroxide solution (50 ml) and this boiled. To the cooled solution was added concentrated nitric acid until the areen precipitate of cobalt(II) oxide dissolved, and then sulphur dioxide was bubbled through to ensure complete conversion of the Co II to Co I. The excess sulphur dioxide was boiled off and the cooled solution 184 II analysed for Co by buffering with sodium acetate and ammonia, and titrating with a 0.1M ethylenediaminetetraacetic acid solution using murexide as indicator. This gave the amount of cobalt as (15.08 + 0.04)%, compared to the stoichiometric value of 15.36%.

12.2.2 THE HOMOGENEOUS RATE

The rate of aquation of bromopentamminecobalt(III) bromide is 2+ conveniently followed spectrophotometrically, for the Co(NH3)5Br ion has an intense charge-transfer peak in the ultra-violet region, which is absent in the Co(NH3)5(1120)3+ ion. The maximum molar absorptivity at 253 nm was determined in a cell of 1 cm path length with a Hitachi Perkin-Elmer model 124 double beam spectrophotometer, 1 to be (1.766 + 0.002) x 104cm-1M , as compared to previous values 4 -1 -1 176b 4 -1 1 185 of: 1.7 4 x 104cm M ;1.7 x 10 cm M ; and 1.862 x 104 -1 -1 186 cm M . 187 The complex is negligibly photoactive. However, the rate constant was determined in the light and dark at 25°C by following the disappearance of reactant spectrophotometrically over approximately 36 hours, using a starting concentration of about 5 x 10-5M in 0.01M perchloric acid. There was no significant difference between the values obtained, which together give the rate constant as (5.79 0.07) -6 -1 176b x 10 s. . The only other value actually determined at 250C and 154

- 6 -1 in acid solution is (6.17 + 0.05) x 10 s The value of Lamb and 188 -6 -1 Marden of 6.51 x 10 s was determined in neutral solution by conductometry, and is consequently expected to be higher, since there is base hydrolysis in addition to the aquation and the platinum conductance electrodes are catalysts. The value attributed177 to 189 -1 Adamson and Basolo of 6.3 x 10 6s at 25°C in fact does not exist. s These workers determined a rate constant of 1.28 x 10 4 -1 at 48°C by 190 spectrophotometry, and Brtinsted and Livingston report a value of -6 -I 1.60 x 10 s at 15°C measured by colorimetry. Interpolating between these last two values, by means of the Arrhenius equation, -1 o gives a rate constant of 6.73 x 10-6 s at 25 C.

The equilibrium constant of reaction (42) with X = Br at 25°C 176b has been determined spectrophotometrically as 0.98 + 0.07M at 2 an ionic strength of 10 M. Other values are 5.8M at zero ionic 191 192 strength and 2.6M at an ionic strength of 0.511.

12.2.3 KINETIC RUNS WITH A FOIL CATALYST 2 A platinum foil substrate (13 cm geometric area) was welded onto a platinum wire, which was sealed into the soda glass end of a graded glass tube. This substrate was cleanpd prior to platinization by dipping it into boiling dilute aqua regia, washing, dipping in warm concentrated nitric acid, washing, cathodizing in 0.1M sulphuric acid -2 at 38 mA cm for 10 minutes, and finally washing again. Platinization was effected from a solution 3.6% (0.074M) in chloroplatinic acid 4 -2 and 0.016% (4.2 x 10 M) in lead acetate at 76.7 mA cm for 8 minutes. A single platinum foil counter electrode (anode) was used, and the working electrode (cathode) was turned around after 4 minutes. The sooty black platinized platinum catalyst was subsequently stored in 1M percnloric acid.

The kinetic runs were performed by weighing the requisite amount of bromopentamminecobalt(III) bromide (ca. 20 mg) into 0.5 4, of 10-211 perchloric acid and stirring, the complex taking about 10 minutes to dissolve, to give a -.4.0-4m solution. Approximately 150 g of this solution was then weighed out and placed in a pyrex beaker, with a ground glass flange and fitted with a perspex lid, in a thermostat. At a time, to , the catalyst was introduced into the solution, the

155

Figure 35. Henry Law kinetic plots at constant stirring.

0.6 T=26°C V0=150 ml

0.4 — s ks 3.9 xi° -6 s

0.2

0 L)

4-z 0 5- 0,6 T5°C Vo=150 mt.

00 0o

0.4. S:Nks=1.0A-X10-6 S

0.2 0 100 200 io3s 156

Figure 36. Henry Law kinetic plot at varied stirring.

Slow iFas SlowStirrin9 :No Stirring T=2 5`C St4-1-4,3 V0=137'711_ Sxlcs= 10.23 x10-g s 0,0

ISAIss=19.36 x10-' s

-0.5

I I

I I.

-1.0 I I 5Ak=303x10 -6 I s 1- 1 I I 1 I 1 I I 0 1 i I I I I t t I I 1 I 1 I 1 I I I 1 I I 1 I I I IO I I I I I I I 1 I I I I I 1 oSXks=2.50x10-6 s 1:1 I 1 I I 1 I t i — ..-- i I 0 100 200

•••••••tc.:_i= ss.,/ 1 /103s ivt 157

latter being continually agitated by means of a magnetic stirrer. Periodically, samples of-approximately 5 ml were removed and analysed spectrophotometrically at 253 nm.

For the kinetic plots, eqn. (62) (Henry's law) can be re-arranged to: n lnA kb(tn -to) = - SXk i E [(t.-t. ] lnA (109 n s 1-11 )/V.-1 o =1 where A is the absorbance. Putting C for the molar absorptivity, n for a path length of unity:

A = ec n n (105) A plot of the left hand side of eqn. (104) against the sigma function should thus give a straight line of slope -SXk s and an intercept at t of 1nA o o. The results for a run at 25C and one at 5°C are shown in Figure 35. As can be seen, at 25°C the points fall on a straight line only after a period of 30 minutes. This indicates that a significant time was required to reach adsorptive equilibrium, perhaps because of diffusion in the pores of the platinized platinum deposit. Support for this idea comes from the run at 5°C, where the points fall on a straight line only after 2 hours. The lnA values expected for o no adsorption are shown as crosses.

Another run at 25°C was carried out with variable stirring. The foil was bent so that it was horizontal, and mixing was achieved with a magnetic stirrer beneath and a paddle stirrer above it. The results are plotted in Figure 36. As can be seen, the rate of reaction increased with stirring, and hence was at least partly diffusion controlled. Consequently, it was decided to use a rotating disc catalyst, where the hydrodynamic stirring conditions are well defined (Chapter 11). The expected value of 1nA0 in Figure 36 is marginally smaller than the extrapolated value. Perhaps such intense stirring assists the adsorptive equilibration processes.

12.3 ROTATING DISC EXPERIMENTS 2 A trumpet shaped RDS of diameter 5.66 cm (geometric area 25.16 cm ) was used. It consisted of a disc of pure platinum 1 mm thick, soldered onto a trumpet shaped stainless steel former. The RDE 158

was detachably mounted onto a motor driven rig as shown schematically in Figure 37. The speed of the d.c. motor was controlled by means of a Servomex M.C.43 unit, and set with reference to a stroboscopic disc. The latter was occasionally checked against a Racal tachometer generator type MA.38 connected to a Racal universal counter type 835.

The sides of the platinum disc and stainless steel former were protected by two coats of Rockhard clear lacquer AR.1501 (Ault and Wiborg Industrial Finishes, Ltd); which is an Araldite epoxy-resin based lacquer. The RDE was a little pitted on the sides, and since the coating process involved curing at 180°C, bubbles persistently formed in the lacquer layer. This problem was eventually overcome by first filling the pit-holes with Araldite adhesive. Then the RDE was coated by immersing it in the lacquer to just below the screw holes and brushing on lacquer above the screw holes, draining for 30 minutes, and baking the coating at 180°C for 30 minutes.

The lacquer on the platinum face was removed by brushing on Stripalene 713 (Sunbeam Anti-Corrosives, Ltd),• which softened it, and wiping with a paper tissue. The disc surface was then gently abraded with 00 grade emery cloth, the latter being placed on a flat black of ground glass.

Because of the hydrodynamic flow in the rotating disc system (RDS), it was necessary to see if effective mixing occurred between the portions of solution beneath and above the disc itself. This was achieved by introducing a small sample of concentrated potassium permanganate solution into the RDS with only water, above, and then below, the disc level. For a 500 ml vessel of diameter 8 cm, mixing was complete in about 40s at 100 r.p.m. and lOs at 1000 r.p.m. For a 1 t, vessel of diameter 10.5 cm, mixing was complete in about 30s at 100 r.p.m. and 5s at 1000 r.p.m. Times of mixing from top to bottom were a little quicker than from bottom to top, as expected from gravitational stability. Hence, with regard to mixing considerations alone, a lower limit of 100 r.p.m. is set for kinetic runs, otherwise the time of mixing introduces a significant error into the measurements. 159

Figure 37. • Rotating disc rig.

Ele.ctri cal Connect ton to Racal Counter

Tachometer Generator

Electrical Connection to Servornex Unit D.C. Motor

Gear Box

Pta.t.inum Wire +---Stroboscopie Disc Cup Mercury Platinum Contact

Hollow Steel ShaFt

Insulated Rod (Screws into Disc)

Hollow steel shaFb •

Nylon Retaining Screw X 3)

Adjustable Table TuF•nol Insulating Sleeve 16o

12.3.1 PLATINIZATION

The platinum face of the disc was briefly abraded with 00 grade emery cloth as described above, and washed. It was electrochemically cleaned in 0.2M percnloric acid by the potential scan sequence given in Chapter 8.1, viz.: rotating at 300 r.p.m., 2s at 1.8 V(RHE) followed by 30s at 1.2 V(RHE), and then in the quiescent solution, 90s at 1.2 V(RHE) followed by lOs at 0.4- V(RHE). Platinization was effected from a solution 3.5% (0.072M) in chloroplatinic acid, 0.005% (1.3 x 10 4M) in lead acetate, and 2M in hydrochloric acid, at a current density of 30 mA -2 cm for 10 minutes. The counter electrode was a lightly chloridized silver-silver chloride sheet, placed horizontally under the RDE. No gas was evolved from this electrode, and the initial chloridization prevented oxidation of metallic silver by the chloroplatinate ions.

During the platinization the RDE was rotated at 200 r.p.m. This was quite critical. At 300 r.p.m. a grey compact deposit was formed at the rim of the RDE only, and nothing in the centre. At 150 r.p.m. a black deposit formed, over the whole surface of the RDE and gas (hydrogen) was evolved. But at 200 r.p.m. a black velvety deposit formed with no gassing. These phenomena are reconciled by Newman's theoretical work outlined in Chapter 11.1.3. With fast stirring the current is below the diffusion limiting value and so the current is concentrated at the edge of the RDE. With slow stirring the current exceeds the diffusion limiting value and hydrogen is evolved. Presumably, for the present system a rotation speed of 200 r.p.m. corresponds approximately to a diffusion limiting current, and hence to a uniform current density distribution. The Levich limiting current density, based on the viscosity of water and the 2- molar conductance of PtC16 at infinite dilution, is 38 mA cm-2.

Prior to use of the silver-silver chloride counter electrode, a platinum foil one was tried, surrounding the RDE. The hydrochloric acid in the plating bath was omitted. At 200 r.p.m. the deposit changed from black at the rim to grey nearer the centre, and nothing was actually deposited at the centre. This must have been due to the counter electrode arrangement and the lack of a high concentration of additional electrolyte (hydrochloric acid), leading to an ohmically more accessible path to the edge of the RDE. 161

12.3.2 AREA DETERMINATION

The real surface area of the RDE was determined from the length of the hydrogen arrest of the galvanostatic charging curve, taken in 1M perchloric acid at 25.0 + 0.1°C. The procedure was that used in Chapter 8.1, and the cell is shown in Figure 38. During the initial nitrogen purge the electrode was rotated at 100 r.p.m., and also for that part of the electrochemical cleaning which involved stirring.

As in Chapter 8.1, the cathodic and anodic charging curves were of similar form and the lengths of the hydrogen arrests (corrected for double layer charging) varied with charging current. The points of intersection of the cathodic and anodic arrest length-current curves were again taken as characterising the area, crossing at 2 between 40 and 100 mA. The figure of 280 pc cm was used once 2 more to convert the data in pc to areas in cm .

The area of the catalyst (RDE) was determined periodically during series of kinetic experiments, to observe the decrease with time. The data are given in Table 16 and plotted in Figure 39. Over the whole period of observation the average rate of decrease of area is 0.26% per day, which is of the order of magnitude expected for ambient temperatures (Chapter 7.3) from other work.

An anomaly apparent in the present work is that the initial roughness factor is about half that expected on the basis of the specific area data in Chapter 8 and the coulombic efficiency data 2 -1 -2 in Chapter Thus from Figure 31, C = cm C at 30 mA cm 5. c 53 -2 for W tkJ 24 C cm . e However, from Table 16, = 456 for Wic = 17.9 C cm-2, so C c = 26 cm C 1. According to eqn. (32) this means that the mass specific area (a) is intrinsically low in high chloride media, or else the coulombic efficiency (Y) is lower, or perhaps a combination of both.

To discover the answer, an experiment was carried out along the lines of those in Chapter 8. The plating solution was 2M in hydrochloric acid and plating was performed at a constant current density -2 of 30 mA cm for 10 minutes, and not at constant potential. The deposition potential rose from roughly 56 mV at the beginning of the platinization to 85 mV after 5 minutes, and thereafter remained 162

Figure 38. Area determination apparatus for rotating disc system.

Nitrogen Inlets

Nitrogen ()tale

Nitro en Outle

osl6a I Level r\ Sintered GlQss Disc Porosity3

Satutroeed Solution of Quinhydrone Percilloric Acid

'-',,Sintered Glass Disc Porosity 3 163

Table 16

The Decrease of Area of a Platinized Platinum Electrode with Time.

Plated from a solution of 3.5% chloroplatinic acid, 0.005% -2 lead acetate and 2M in hydrochloric acid at 30 mA cm for 2 10 minutes. The electrode (A = 25.16 cm ) was prepared on day 0.

2 ' . Ila Real Area/cm Roughness Factor 1 11500 456 17 9900 395 62 9200 365 136 750o 298

164

Figure 39. Decrease of roughness factor of a platinized platinum rotating disc electrode with Time at Room Temperature (ca. 23°C).

400

50 Time/days 165

2 approximately constant. -The results were: S = 5950 cm 85 cm2mg-1; W = 5.4 mg cm-2, Y = 58%. From eqn. (32), kS = 458, um = m -1 these data lead to G = 26 cm2C . 1 as required. Now, Figure 30 gives, -2 2 -1 for an average current density of 30 mA cm , G = 122 cm mg and m -2 Y = 87% for m=10 mg cm There are indications in Chapter 8 -2 that C is independent of w at least in the range mg cm . m M 5 - 10 Thus, with the high chloride plating solution, the halving of the expected roughness factor was due to a fall of mass specific area by a factor of 0.7 and of coulombic efficiency by a factor of 0067.

Nevertheless, this still leaves the disagreement between the present value of coulombic efficiency of 58% and that of Chapter 5 of 100%. The only difference in preparation was that of the counter electrode and plating cell. There was a silver-silver chloride electrode and a one compartment cell in Chapter 5, and two platinum electrodes and a three compartment cell for tnis'Chapter. It would be rather surprising for such a minor change to cause such a large discrepancy in the coulombic efficiencies. Moreover, the present potential of platinization agrees well with the upper three curves of Figure 4, and the appropriate coulombic efficiency of Chapter 5 was determined with the electrode corresponding to the lower curve of this figure. Thus, perhaps the value of coulombic efficiency of Chapter 5 (100%) is anomalous.

12.3.3 CATALYSIS RUNS

Except for two initial experiments, the catalyst (RISE) was first electrochemically cleaned as in the previous section. The kinetic runs were carried out by making up a solution of bromopentamminecobalt(III) bromide in 0.O1M perchloric acid more concentrated than required. The disc was rotated in a known volume of 0.0I1! perchloric acid thermostatted at 25°C, and at a time t a standard volume of the concentrated o reaction solution was added, so as to give 200 ml of the required concentration. Runs were generally carried out over a period of about 8 hours, and at both 400 and 1000 r.p.m. to check for diffusion control. The reaction rig is shown schematically in Figure 40. The potential was followed, from before addition of the reaction solution, against a quinhydrone electrode in 0.01M perchloric acid, with 0.O1M 166

Figure 40. Apparatus for kinetic runs with the rotating disc.

Counter Electrocie ;For Runs ab Consont Poter&ial Only

crey, TVIread ToInt Somplins Port

10211 Percbtoric 10-1M Perchtbric Acid Thermostab Acid Level

r Sintered Glass Disc Par osity3 Reaction Satution Beckman r:. Porous Sa burated Solutltr% Plus o Quinhydrone in 10-111 Perchloric Acid

, Sintered Class Disc Porosity 2 167

perchloric acid in the Luggin capillary. For two runs the catalyst was held at constant potential, using a 0.01M perchloric acid bridge connection to a platinum foil counter electrode (Figure 40). The junction with the reaction solution was through a ceramic porous plug, derived from a Beckman remote junction glass electrode (150674/RRLB). Periodically samples of 1 ml were removed and analysed spectrophotometrically at 300 nm in special semi-micro cells. The latter had thick side walls made of black quartz, and were manufactured by Hellma Ltd. At 300 nm the molar absorptivity was measured to be -1 -1 176b -1 -1 973 + 9 cm M , compared to a previous value of 987 cm M . The homogeneous reaction was also followed, so that the actual starting concentration of the heterogeneous run could be determined from interpolation.

The measured data were plotted according to eqn. (104), and the results are given in Table 17. The scatter was. generally greater than with a first order plot, presumably because of the cumulative errors inherent in the sigma function and homogeneous correction. At the higher concentrations t OD 3 Mi, the heterogeneous contribution is very low, and the scatter in the plots becomes very large, due to the allowance for the homogeneous reaction. For these cases it is better to modify eqn. (104) by omitting the kb(tn-to) term. The slope is then for the overall reaction, and can be adjusted by subtracting k V b o to give a good approximation to SXk . The real surface areas for Runs 1-11 in Table 17 are roughly average values, while those for Runs 12-21 are interpolated from Figure 39. The variation of potential of the catalyst with time followed the typical curve of Figure 41. The potential (versus NHE unless stated otherwise) before the addition of reactant was about 790-800 mV for Runs 1 and 2 (no pretreatment) and for the rest 940 + 20 mV except for Run 14 (under nitrogen) when it was approximately 890 mV. The potentials E . max listed in Table 17 are the maximum values attained (see Figure 41), and the total rise in potential was between 100 and 200 mV. For Runs 16 and 17 the potential was held constant; the current decreasing from 400 to 100 pA in Run 16 and from 900 to 100 pA in Run 17, over the measurement periods of roughly 7 hours. It should be noted that at the lower potential of Run. 16 the catalysed 168

Table 17

Kinetic and Adsorption Results for Heterogeneous Catalysis from the Henry Law Treatment.

-4Ma -/ -4Mb S/cm2 Run No. co/10 c/10 uVr.p.m. EXks/ c /10 1° Emax b -1 10 ts mol cm.'" mV(NHE) c 1 5.20 4.6o woo 1.14 1.46 934 c 2 5.19 4.60 400 1.11 -0.95 896 3 4.92 3.88 400 3.19 8.90 990 -6700 4 4.85 3.97 1000 2.87 7.32 1009 5 5.01 3.88 400 2.98 5.89 1004' 6 4.55 4.13 400 1.05 7.98 1062 7 5.04 4.50 low 0.63 3.67 1065 8 0.74 0.67 400 4.97 8.36 1052 -6800 9 0.80 0.68 Imo 5.46 '7.50 1051 10 12.7 12.3 400 -0 39.6 1065 11 12.2 11.9 1000 -0 32.2 1059 12 4.65 4.30 400 1.35 12.9 1071 10100 13 4.79 4.16 1000 1.86 10.4 1076 wood 14d 4.36 4.06 1000 1.05 12.2 1080 9800 15 4.65 4.39 400 1.11 12.5 1030 8800 16elf 5.09 3.29 400 17.6 3.09 980 8800 17elf 5.07 4.83 400 -0 6.53 1130 870o 18f 4.81 3.19 400 o.58 16.6 1083 870o 19fo 5.06 4.20 4-00 1.09 7.98 - 8600 f 20 12.0 7.19 400 0.04 27.3 1070 8000 f h 21 9 5.04 4.37 400 0.52 8.1+1 1078 7550

a This is the extrapolated (equilibrium) value of starting concentration. b This is the average of the actual starting concentration and the final concentration measured. c No pretreatment of catalyst surface. d Run under nitrogen. e Potential held constant during the run. f Cobalt(II) tested lor (see Table 18). Only two readings taken. g 176c): h Different pretreatment (after Archer at 100 r.p.m., 15 minutes at 0.4 V(NHE) and 10 minutes at 1.11 V(NHE).

169

Figure 41. Variation of potential of the rotating disc during a kinetic run.

1.05 forti.g•••

1.00 LU

0.95

0 10 20 Vio3s 170

rate of disappearance of the complex is greatly enhanced, whereas at the higher potential of Run 17 it is completely suppressed. This is in qualitative agreement with the result of Archer.176c.

12.3.4 PRODUCTION OF COBALT(II) 176b After Runs 16-21, the reaction mixture was analysed for the presence of cobalt(II), by complexation with chloride and spectrophoto- metric estimation. A small aliquot (5 ml) of the solution was made up to 25 ml with concentrated hydrochloric acid and the spectrum measured in the range 600 - 750 nm. For this the Hitachi Perkin-Elmer model 124 instrument was used with the 165 recorder, the latter being set on a full scale deflection of ten times the normal, thus effectively acting as a scale expansion accessory of X10. The cobalt(II) concentration was estimated-176b at 690 nm where C is 450, giving a lower 6 limit of detection of approximately 2 x 10 m. In all cases, except Run 17, the presence of cobalt(II) was detected, usually in amounts corresponding roughly to those expected if all the enhanced reaction (over hnd above the homogeneous reaction) had proceeded by reduction of the cobalt(III) complex. With Run 17, of course, no cobalt(II) was to be expected since the heterogeneous reaction was completely suppressed. The results are collected in Table 18.

If the enhanced reaction were due entirely to reduction, c(E) (see Table 18) would equal c(II); and if there were at least some genuine catalysis of the aquation process without reduction, c(E) would be greater than c(II). The latter inequality is true only for Runs 17, 19 and 21, though it is doubtful whether conclusions about the presence of catalysis can be drawn from these rather marginal differences. What is clear is that the major part, at least, of the enhanced reaction is due to the decomposition of the complex to a cobalt(II) species. This is emphasized by Runs 16 and 17, where disappearance of the reactant is very much faster with cathodization (Run 16) and near to the homogeneous rate with anodization (Run 17). To confirm this reduction to Co(II), a run at high concentration (Run 20) over a long period was performed, and the supporting acid analysed afterwards. Decomposition of the complex should liberate ammonia and neutralise some of the acid. The acid was analysed before and after the run with sodium hydroxide solution, 171

Table 18

Results of Analyses for Cobalt(II)

Run Reaction ci/10 4M cf/10 4M c(PH)/10-4M c(E)/10-4M c(II)/10 4M No. time/hours 16 6.32 5.23 0.62 0.64 3.97 4.00 17 8.35 5.35 4.32 0.86 0.17 0000 18 52.45 5.53 0.85 3.68 1.00 1.45 19 13.00 5.40 3.00 1.27 1.13 1.00 20 108.25 13.08 1.30 11.71 0.07 3.20 21 13.00 5.36 3.39 1.26 0.71 0.65 c. = Starting concentration of [Co(NH3 )513r]Br2 (calculated from homogeneous run). c f = Final concentration of [Co(NH3 )5Br]Br2. c(PH)= Concentration of product that would be produced by homogeneous reaction alone (calculated). c(E) = ic. - c f - c(PH), is a measure of the total enhancement of the reaction. c(II)= Concentration of cobalt(II) at end of run. 172

using bromocresol purple as indicator. The determination of acid was done quickly, to minimize hydrolysis of the complex. The acidity decreased from an original value of 0.00966M to 0000885M; for -4 II 3.20 x 10 M of Co there should have been the equivalent of 16 x 10 4M ammonia released, compared to the measured decrease of acidity of 8.1 x 10-4M. Altnough the figures differ by a factor of 2, at such low levels of acid concentration the facts support the decomposition conclusion.

In Runs 18 and 20 in Table 18, c(II) is significantly greater than c(E). This implies reduction of tne aquation product ion to cobalt(II). To see if this ion was in fact reduced, the RDE was rotated at 400 r.p.m. in 200 ml of a 1.30 x 10-3M solution of aquopentamininecobalt(III) perchlorate for 89.5 hours, and the solution then analysed for cobalt(II). During the run the potential rose from 941 to 974 mV. No cobalt(II) was detected, nor was there any decrease in acidity of the 10-2M perchloric acid supporting electrolyte. One inference is that the reduction involves bromide ions, and evidence supporting this will be presented in the following section.

12.4 DISCUSSION

That the values of SXk in Table 17 are not constant with varying s concentration demonstrates that the system does not obey Henry's Law adsorption. They do not change in a regular manner with speed of rotation so that the heterogeneous reaction is surface- and not transport-controlled. The complication of parallel reactions of aquation and reductive decomposition does not make it worthwhile to plot alternative functions. Instead, the values of SAk can be re- s interpreted as empirical first order rate constants for the heterogeneous reaction, corrected for the homogeneous aquation and volume changes due to sampling. They are then on the same basis as the kh of Archer and Spiro (see Chapter 11.2.1). Note that they decrease as c increases, 176c in qualitative agreement with Archer's results which were interpreted on a Langmuir model.

From Table 17 the values of surface concentration of the adsorbed reactant, c s, show a very large scatter and are useless for estimating an adsorption isotherm. The value for Run 2 is even negatiye. 173

176c Moreover, most of them exceed the c value found for mercuric m -10 sulphide and silver bromide of 4.2 x 10 mol cm-2. Presumably, these irreproducible effects are a consequence of the parallel reaction system and the pore structure of the catalyst.

The main question arising from the work described in this chapter concerns the agent that reduced the complex. In fact, 2+ the complex ion Co(NH3 )5Br is not thermodynamically stable with respect to reduction in acid aqueous solution. The driving force of the reaction is essentially the protonation of the ammonia ligands. Moreover, uncomplexed cobalt(III) is not stable in aqueous solution196 as can be seen from the couple:

Co34. + e- 72: Co2+; E° = + I.808V; GIG° = 174.4 kJ mol-1 (106)

From the (average) values of stability constants193 the standard Gibbs function changes can be calculated as follows: + NH3 + H NH4+; log K = 9.233; = - 52.7 kJ mol-1 (107)

2+ 2+ Co (aq)+ 5NH3 +H20 - Co(NH3)51120 ; log K = 5.155; be. -29.41a mol-1 (108)

Co3+(aq)+ 5NH3+1120 Co(NH3)0203+; log K = 32.82; d3°= "187.4kJ mol-1(109)

Combining reactions (1.06) + (108) - (109) gives: o -1 o Co(NH3 )5112034+ e Co(NH3)0202 ; tIG *16.4 kJ mol ; E = +0.170V (110)

This standard potential is substantially different from Yalman's194 values of +0.37 and +0.33V. The former was calculated as for reaction (110) only with different and older data; the latter was calculated from experimentally determined equilibrium constants for the reactions: 2+ 2+ - 2Co(NH3)5H20 +13 2Co(NH3 )I +I + 2H20 (111)

3+ _r- Co(NH3)5 ..u2 n -,tr-tmu )/51 T2 +. H2 O (112) together with the redox potential of the iodine-iodide couple. Adding reaction (110) to the aquation equilibrium (using the Gibbs . function change of Mori and Tsuchiya191):

Co(NH3 )5Br2+ + H20 - Co(NH3 )511203+ + Br; = - 4.3 kJ mol (113)

leads to: 174

2+ Co(NH3)5Br24 + H2 O + e Co(NH3)5H20 + Br ; = -20.7 kJ mol;

Eo = +0.215V (114)

From the reaction sequence (114) +.5(107) - (108):

Co(NH3)5Br2+ + 5H++ e- Co2++ 5NH4++ Br-; Mc) = -254.8 kJ mo1-1; E° = +2.641V (115)

This couple can, from the thermodynamic viewpoint, easily oxidise the solvent:

02 + 4H+ + 4e 2H2 0; E° = +1.229V; iNg° = -474.3 kJ mo1-1 (116)

Combining reactions (115) - i-(116) gives:. 2+ + 2+ + Co(NH3)5Br + 4H +2H20 z•-. Co + 5NH4 + Br + t02 ; Lie = -136.5 kJ mo1-1 (117)

Moreover, the starting complex could spontaneously disproportionate in acid solution) as the bromine-bromide couple is less noble:

Br2 + 2e 2Br-; E° = 1.087V; = -209.8 kJ mol-1 (118)

Using reactions (115) - 1(118): 2+ + 2+ + o -1 Co(NH3)5Br + 5H 4---" Co + 5NH4 + 2Br2 ; C.13 = -149.9 kJ mol (119)

Thus there are two possible paths for the decomposition of the starting complex, namely reactions (117) and (119), although there is no decomposition in the homogeneous state. Reaction (119) has circumstantial evidence behind it, one aspect of which has already been mentioned in the previous section. The bromine-bromide couple is reversible and cobalt redox couples are generally irreversible, so according to the theory outlined in Chapter 10.1, if there is electron exchange catalysis, the potential should be near to the bromine-bromide couple value. Table 17 shows this to be so, the potential jumping upwards by 100 - 200 mV to reach an average value of 1.064 + 0.011 V (Figure 41).

Another possibility is the oxidation of the platinum metal catalyst. In the presence of bromide ions: 2- -1 PtBr4 + 2e = Pt + 4Br; E° = +0.581V; - -112.1 kJ mol (120) so with reaction (115) this gives: 2+ + 5H+ / Co(NH3 )5Br + 2Pt + Br Co2++ 511N++ 2PtBr42-'; b3° = -198.7 kJ mo1-1 (121) 175

Indeed, platinum ions have been produced in 3.6M hydrochloric acid195 66 by anodic polarization, and in 1M sulphuric acid by cyclic voltammetry (see Chapter 3.4).

12.5 CONCLUSION

The heterogeneous catalysis by platinized platinum of the disappearance of the bromopentamminecobalt(III) ion in acid solution has been found to be due largely to reductive decomposition. The present work is in qualitative agreement with that using a smooth platinum catalyst, but whether a similar heterogeneous process occurs on the latter is not known. Work on this aspect is currently being carried out in this laboratory by Dr R.J. Mureinik. 176

REFERENCES

1. N.D. Zelinsky and M.B. Turowa-Pollak, Ber. Deutsch. chem. Ges., 1925, 58, 1298. 2. 0. Lummer and F. Kurlbaum, Sitzungsber. Kgl. Preuss. Akad. Wiss., 1894, 229. 3. A.A. Balandin, V.I. Spitsyn, L.I. Barsova, and V.I. Duzhenkov; Russ. J. Phys. Chem., 1959, 63, 17 (Abstracts). 4. 0. Lummer and F. Kurlbaum, Ann. Phys. Chem., 1892, 46, 204. 5. F. Kurlbaum and O. Lummer, Vern. phys. Ges. Berlin, 1895, 14, 56. 6. F. Kohlrausch, Ann. Phys. Chem., 1897, 60, 315. 7- G. Jones and D.M. Bollinger, J. Amer. Chem. Soc., 1935, 7, 280. 8. F. Kohlrausch and L. Holborn, "Das Leitverm5gen der Elektrolyte", B.G. Teubner, Leipzig, 1898, p.9. 9.. F. Kohlrausch, Ann. Phys. Chem., 1897, la, 423. 10. A.I. Vogel, "A Textbook of Macro and Semimicro Qualitative Inorganic Analysis", 4th ed., Longmans, London, 1954, p.581. 11. "Platinum", International Nickel Limited, London, p.11. 12. G. Hevesy and T. Somiya, Z. phys. Chem., 1934, 171, 41. 13. A. Slygin and A. Frumkin, Acta physicochim. URSS, 1935, 2, 791. 14. R.N. Goldberg and•L.G. Hepler, Chem. Rev., 1968, 68, 229. 15. V.M. Latimer, "Oxidation Potentials", 2nd ed., Prentice-Hall, Englewood Cliffs, N.J., 1952, (a) p.204; (b) p.343, (c) p.341. 16. E. Sonstadt, Proc. Chem. Soc., 1898, 14, (a) 25, (b) 179. 17. E.H. Archibald, J. Chem. Soc., 1920, 1104. 18. A.A. Grinberg and Yu. N. Kukushkin, Russ. J. Inorg. Chem., 1961, 6, 554; A.D. Westland and L. Westland, Proc. 7th Int. Conf. Co-ord. Chem., (Abstracts), 1962, p.251. 19. R. Dreyer, I. Dreyer, and D. Rettig, Z. phys. Chem. (Leipzig), 1963, 224, 199. 20. N.M. Nikolaeva, B.V. Ptitsyn and E.D. Pastukhova, Russ. J. Inorg. Chem., 1965, 10, 574. 21. A.T. Hubbard and F.C. Anson, Anal. Chem., 1966, 78, 1887. 22. G. Grube and H. Reinhardt, Z. Elektrochem., 1931, E, 307. 23. O. Stelling, ibid., 1931, 27_, 321. 24. H. Yamamoto, Y. Kobaku, T. Nagai and T. Takei, Denki Kagaku, 1965, 804; Chem. Abstr., 1966, 64, 13747h. 25. C. Bernard, Electrochini. Acta, 1970, 15, 271. 177

26. A.T. Hubbard, J. Electroanal. Chem., 1969, 22, 165. 27. A.L.Y. Lau and A.T. Hubbard, ibid., 1970, 24, 237. 28. A.N. Frumkin,' Trans. Faraday Soc., 1959, 22, 156. 29. V.I. Kravtsov and B.V. Simakov, Electrochim. Acta, 1964, 9, 1211. 30. J.J. Lingane, J. Electroanal. Chem., 1964, 2, 94. 31. G.E. Adams, R.B. Bosziewicz and B.D. Michael, Trans. Faraday Soc., 1968, 64, 1256. 32. A.S. Ghosh-Mazumdar and E.J. Hart, Int. J. Radiat. Phys. Chem., 1969, 1, 165. 33.M. Fleischmann and H.R. Thirsk, Advan. Electrochem. Electrochem. Eng., 1963, 123. 34. K.M. Gorbunova and Yu. H. Polukarov, Advan. Electrochem. Electrochem. Eng., 1967, 2, 249 (see p.263). 35. P. Delahay, "Double Layer and Electrode Kinetics", Interscience, New York, 1965, p.128. 36. H.A. Roth and W.R. Lasko, Amer. Chem. Soc., Div. Fuel Chem., Prepr., 1967, 11, 241. 37.S. Toschev and B. Mutaftschiew, Electrochim. Acta, 1964, 9, 1203. 38. F.A. Cotton and G,. Wilkinson, "Advanced Inorganic Chemistry", 2nd ed., Interscience, New York, 1966, p.350. 39. J.P. Hoare, "The Electrochemistry of Oxygen", Interscience, New York, 1968, (a) Chapter II, (b) pp.171-175. 40. G.J. Hills and D.J.G. Ives in "Reference Electrodes", D.J.G. Ives and G.J. Janz, Ed., Academic Press, New York, 1961, pp.106-108. 41. F.C. Anson, Anal. Chem., 1959, .311 1451. 42. W.D. Cooper and R. Parsons, Trans. Faraday Soc., 1970, 66, 1698. 43.S. Popoff, A.H. Kunz and R.D. Snow, J. Phys. Chem., 1928, -42, 1056. 44. N.J. Joncich and N. nackerman, J. Electrochem. Soc., 1964, 111, 1286. 45. G.J. Hills and D.J.G. Ives, J. Chem. Soc., 1951, 305. 46. R. Thacker in "Hydrocarbon Fuel Cell Technology", B.S. Baker, Ed., Academic Press, New York, 1965, p.525. 47. R. Woods, Electrochim.' Acta, 1968, 12, 1967. 48. R.H. Atkinson, Trans. Inst. Metal Finish., 1958-59, ,x, 7. 49. K.S. Indira, S. Sampath and K.S.G. Doss, Metal Finish., 1969, 44; abstracted in Plat. Met. Rev., 1969, 12, 156. 50. L.P. Hammett and A.E. Lorch, J. Amer. Chem. Soc., 1933, j2, 70. 178

51. H.T. Beans and L.P. Hammett, ibid., 1925, 1215. 52. F.H. Reid, Metallurg: Rev. (London), 1963, 8, 167. 53. W.M. MacNevin and M. Levitsky, Anal. Chem., 1952, 24, 973. 54. G. Bianchi, Ann. Chim. (Rome), 1950, 40, 222. 55. F. Kurlbaum, Ann. Phys. Chem., 1899, j2a, 846. 56. E.C. Potter, "Electrochemistry", Cleaver-Hume Press, London, 1956, p.102. 57. C. Bernard, Electrochim. Acta, 1969, 14, 143. 38. R.R. Giles, J.A. Harrison and H.R. Thirsk, J. Electroanal. Chem., 1969, 20, 47. 59. J.A. Harrison, personal communication, 1970. 60. J.O'M. Bockris and A. Damjanovic, Mod. Aspects Electrochem., 1964, 2, 224. 61. D.F.A. Koch, Extended Abstracts, 131st National Meeting of the Electrochemical Society, May 1967, p.32, No.174. 62. I.R. Jonasson, Ph.D. Thesis, Adelaide, 1967, p.148. 63. A.A. Balandin, V.I. Spitsyn, L.I. Barsova, A.B. Agronomov and N.P. Dobrosel'skaya, Russ. J. Phys. Chem., 1967, 41, 1420. 64. M. Bonnemay, Z. Elektrochem., 1955, 21, 798. 63. F.C. Anson and D.M. King, Anal. Chem., 1962, :/L., 362. 66. T. Biegler, J. Blectrochem. Soc., 1969, 116, 1131. 67. A. Bewick and A.M. Tuxford, personal communication, 1969. 68. A. Hickling, Trans. Faraday Soc., 1945, 41, 333. 69. S. Gilman, J. Electroanal. Chem., 1965, 2, 276. 70. S.D. James, J. Electrochem. Soc., 1969, 116, 1681. 71. M. Hansen, "Constitution of Binary Alloys, 2nd ed., McGraw-Hill, London, 1958, p.1096. 72. N.A. Puschin and P.N. Laschtschenko, Z. anorg. Chem., 1909, 62, 34. 73. "Selected Values of Chemical Tnermodynamic Properties", N.B.S. Circular 500, Washington D.C., 1952, pp.9, 158. 74. "Analytical Methods for Atomic Absorption Spectroscopy", Perkin-Elmer, Norwalk, Connecticut, 1968. 75. D.R. Crow and J.V. Westwood, "Polarograpny", Methuen, London, 1968, Chaps. 1-4 and 8. 76. B. Breyer and H.H. Bauer, "Alternating Current Polarograpny and Tensammetry", Interscience, New York, 1963, Chap. 2. 77. J.E.B. Randles in E. Yeager, Ed., "Transactions of the Symposium on Electrode Processes.", Wiley, New York, 1961, p.209. 179

78. J.01 M. Bockris and A.K.M.S. Hug, Proc. Roy. Soc. A,1956, 237, 277. 79. "Interatomic Distances Supplement", Chem. Soc. Special Publ. No.18, London, 1965, p. S 10s. 80. J.P. Hoare, Electrocnim. Acta, 1964, 9, 599. 81. A.N. Frumkin, Advan. Electrochem. Electrocnem. Eng., 1963, 3_, 287. 82. S. Gilman, Advan. Electroanal. 0nem., 1967, 2, 111. 83. T. Biegler, D.A.J. Rand, and R. Woods, J. Electroanal. Chem., 1971, 29, 269. 84. A. Damjanovic, Mod. Aspects Electrochem., 1969, 5., 369.

85. J.D.E. McIntyre and D.M. Kolb, Symp. Faraday Soc., 1970, 99. 86. M.A. Barrett and R. Parsons, ibid., 1970, 4, 72. 87. A. Bewick and A.M. Tuxford, ibid., 1970, 4, 114. 88. T. Biegler and R. Woods, J. Electroanal. Chem., 1969, 20, 73. 89. T.B. Warner and S. Schuldiner, J. Electrochem. Soc., 1965, 112, 853. 90.D. Gilroy and B.E. Conway, Can. J. Chem., 1968, 46, 875. 91. J.P. Hoare, J. Electrochem. Soc., 1969, 116, 612, 1390. 92. A.K.N. Reddy, M.A. Genshaw and J.O'M. Bockris, J. Chem. Phys., 1968, 48, 671. 93. R. Greef, ibid., 1969, a, 3148. 94. M.A. Genshaw and J.O'M. Bockris, ibid., 1969, a, 3149. 95. J.L. Ord and F.C. Ho, J. Electrochem. Chem., 1971, 118, 46. 96. B. Ernsler, Discuss. Faraday Soc., 1947, 1, 269. 97. F.P. Bowden and E.K. Rideal, Proc. Roy. Soc. A, 1928, 120, 59. 98.A.I. Pletyushkina, G.P. Knomcnenko and G.D. Vovchenko in "Kataliticheskie Reaktsii v Zhidkoi Faze", Akademiya Nauk Kazakhskoi SSR, Kazaknskii Gosudarstvennyi Konferentsii, Alma-Ata, 1963, p.295. 99. G.A. Bogdanovskii, G.P. Knomchenko and G.D. Vovcnenko, Russ. J. Pnys. Chem., 1965,,9, 747. 100.S.B. Brummer, J. Phys. Chem., 1965, 69, 562. 101.J. Giner and J.M. Parry, U.S. Clearing House Fed. Sci. Tech. Inform., AD 648727, 1967, pp.82-84. 102.J. Giner, J.M. Parry, S. Smith and M. Turcnan, J. Electrocnem. Soc., 1969, 116, 1692. 103.S.D. Drummer, J.I. Ford and M.J. Turner, J. Phys. Chem., 1965, 69, 3424. 180

104.S. Gilman, Electrochim. Acta, 1964, 9, 1025. 105.J.S. Mayell and S.H. Langer, J. Electrochem. Soc., 1964, 111, 438. 106.S. Schuldiner and R.M. Roe, ibid., 1963, 110, 332. 107.R.J. Brodd and N. Hackerman, ibid., 1957, 104, 704. 108.J.J. McMullen and N. Hackerman, ibid., 1959, 106, 341. 109.F.S. Feates, D.J.G. Ives and J.H. Pryor, ibid., 1956, 103, 580. 110.M.W. Breiter, J. Electroanal. Chem., 1964, 2, 38. 111.C. Bernard and J. Clavilier, C.R. Acad. Sdi. C, 1966, 2, 1340. 112.M. Prigent, Rev. Gen. Elec., 1965, .711, 69. 113.V.I. Podlovchenko and R.P. Petukhova, Sov. Electrochem., 1969, 5_, 349; 1970, 6, 185. 114.I.R. Jonasson and D.R. Stranks, Electrochim. Acta, 1968, 1.2, 1147. 115.J.M. Blakely and H. Mykura, Acta Met., 1962, 10, 565. 116.A.J. Melmed, J. Appl. Phys., 1967, 24ii, 1885. 117.R. Thacker, Nature, 1966, 212, 182. 118.F. Nagy, Gy. Horanyi and Gy. Vertes, Acta Chim. Acad. Sci. Hung., 1962, .11-1, 35. 119.C.N. Myers and S.F. Acree, Amer. Chem. J., 1913, 50, 396. 120.A. Frumkin and A. Slygin, Acta physicochim. URSS, 1936, 5, 819. 121.R.A. Robinson and R.H. Stokes, "Electrolyte Solutions", 2nd ed., Butterworthsf London, 1965, (a) p.463, (b) p.491. 122.A.V. Babaeva, Bull. Acad. Sci. URSS, C1.Sci.Chim., 1943, 171. 123.C.K. JArgensen, Acta Chem. Scand., 1956, 10, 518. 124.J. Lewin, J. Amer. Chem. Soc., 1951, 22, 3906. 125.J. Chat, G.A. Gamlen, and L.E. Orgel, J. Chem. Soc., 1958, 486. 126.J.R. Joy and M. Orchin? J. Amer. Chem. Soc., 1959, 81, 305. 127.G.J. Janz and D.J.G. Ives, in "Reference Electrodes", D.J.G. Ives and G.J. Janz, Eds., Academic Press, New York, 1961, Chap. 6. 128.I.M. Klotz, "Some Principles of Energetics in Biochemical Reactions", Academic Press, New York, 1957, Chap. V. 129.G.W. Tindall and S. Bruckenstein, Electrochim. Acta, 1971, 16, 245. 130.A.H. Taylor, S. Kirkland, and S.B. Brummer, Trans. Faraday Soc., 1971, 67, 809. 131.D.O. Hayward and R. Chappell, personal communication, 1971. 132.D.O. Hayward and B.M.W. Trapnell, "Chemisorption", 2nd ed., Butterworths, London, 1964, p.300. 181

133.S. Nagami, J. Chem. Soc. Japan, 1927, 48, 501; Chem. Abstr., 1928, 22, 2097. 134.F. Daniels, J.H. Mathews, J.W. Williams, P. Bender, and R.A. Alberty, "Experimental Physical Chemistry", 5th ed., McGraw-Hill, New York, 1956, p.177. 135.D.P. Shoemaker and C.W. Garland, "Experiments in Physical Chemistry", McGraw-Hill, New York, 1962, p.444. 136.A. Findlay and J.A. Kitchener, "Practical Physical Chemistry", 8th ed., Longmans, London, 1954, (a) p.206, (b) p.260. 137.A.M. James, "Practical Physical Chemistry", 2nd ed., Churchill, London, 1967, p.45. 138.R.G. Bates, "Determination of pH", Wiley, New York, 1964, p.242. 139.R. Woods, personal communication, 1970. 140.M. Spiro, J.N.Z. Inst. Chem., 1971, 22, 39. 141.M. Spiro and A.B. RavnB, J. Chem. Soc., 1965, 78. 142.M. Spiro and P.W. Griffin, Chem. Comm., 1969, 262. 143.F. Basolo and R.G. Pearson, "Mechanisms of Inorganic Reactions"., 2nd ed., Wiley, New York, 1967, (a) p.330, (b) Chap. 3. 144.F.P. Dwyer and A.M. Sargeson, Nature, 1960, 111, 1022. 145.H.E. Swift and B.E. Douglas, J. Inorg. Nucl. Chem., 1964, 26, 601. 146.W.C. Erdman, H.E. Swift and B.E. Douglas, ibid., 1962, 24, 1365. 147.D. Ben and W.C. Fernelius, ibid., 1959, 10, 269. 148.A. Jensen, F. "Basolo, and H.M. Newmann, J. Amer. Chem. Soc., 1958, 80, 2354. 149.P.S. Walton and M. Spiro, J. Chem. Soc. (B), 1969, 42. 150.J.M. Austin, 0.D.E.:S. Ibrahim, and M. Spiro, ibid., 1969, 669. 151.P.S. Walton, Ph.D. Thesis, University of London, 1967, p.12. 152.M.D. Archer and M. Spiro, J. Chem. Soc. (A), 1970, 68. 153.Idem., ibid., 1970, 73. 154.Idem., ibid., 1970, 78. 155.Idem., ibid., 1970, 82. 156.E.H. Cordes and R.B. Dunlap, AcC. Chem. Res., 1969, 2, 329. 157.E.J. Fendler and J.H. Fendler, Advan. Phys. Org. Chem., 1970, 8, 271. 158.H. Morawetz, Advan. Catalysis, 1969, 20, 341. 159.L.L. Bircumshaw and A.C. Riddiford, Quart. Rev., 1952, 6, 157. 160.V.G. Levich, "Physicochemical Hydrodynamics", Prentice-Hall, Englewood Cliffs, N.J., 1962, Chaps. I and II. 182

161.A.C. Riddiford, Advan. Electrochem. Electrochem. Eng., 1966, 4, 47. 162.J. Newman, ibid., 1967, 5, 87. 163.D.P. Gregory and A.C. Riddiford, J. Chem. Soc., 1956, 3756. 164.J. Newman, J. Phys. Chem., 1966, 22, 1327. 165.W.H. Smyrl and J. Newman, J. Electrochem. Soc., 1971, 118, 1079. 166.J. Newman, ibid., 1966, 113, 1235. 167.V. Maranathe and J. Newman, ibid., 1969, 116, 1704. 168.S. Bruckenstein and B. Miller, ibid., 1970, 117, 1044. 169.W.J. Albery and M.L. Hitchman, Trans. Faraday Soc., 1971, 2408. 170.R.R.M. Johnston, Ph.D. Thesis, University of London, 1965, Chap. II. 171.L. Fleischer and H. Schuberth, Chem. Tech. (Berlin), 1969, 21, 84. 172.E.A. Aykazyan and A.I. Fedorova, Doklady Akad. Nauk. SSSR, 1952, 86, 1137; cited in Ref. 160, p.312. 173.D.H. Angell, T. Dickinson, and R. Greef, Electrochim. Acta, 1968, II, 120. 174.R.S. Bradley, Trans. Faraday Soc., 1938, 7;iLl 278. 175.D.H. Everett and R.H. Ottewill, Eds., "Surf4ce Area Determination", Butterworths, London, 1970. 176.M.D. Archer, Ph.D. Thesis, University of London, 1968, (a) pp.127-133, (b) Chap. V, (c) Chap. X. 177.C.H. Langford and H.B. Gray, "Ligand Substitution Processes", Benjamin, New York, 1965, Chap. 3. 178.W.E. Jones and T.W. Swaddle, Chem. Comm., 1969, 998. 179.D.L. Gay and R. Nalepa, Can. J. Chem., 1970, 48, 910. 180.N.A. Balashova and V.E. Kazarinov, Advan. Electroanal. Chem., 1969, 21 135. 181.S.M. Ifasany, Ph.D. Thesis, University of Adelaide, 1971, pp. 83, 87, 135 and 137. 182.H. Diehl, H. Clark, and H.H. Willard in "Inorganic Syntheses", H.S. Booth, Ed., McGraw-Hill, New York, Vol. I, 1939, p.186. 183.A.I. Vogel, "Quantitative Inorganic Analysis", 3rd ed., Longmans, London, 1962, p.950. 184.F.J. Welcher, "The Analytical Uses of Ethylenediaminetetraacetic Acid", van Nostrand, Princeton, N.J., 1958, p.230. 185.A.R. Olson and T.R. Simonson, J. Chem. Phys., 1949, 12, 1167. 186.M. Linhard and M. Weigel, Z. anorg. allgem. Chem., 1951, 266, 53. 187.A.W. Adamson, Co-ord. Chem. Rev., 1968, 169. 183

188.A.B. Lamb and J.W. Harden, J. Amer. Chem. Soc., 1911, 21, 1873. 189.A.W. Adamson and F. Basolo, Acta Chem. Scand., 1955, 9, 1261, 190.J.N. Br8nsted and R. Livingston, J. Amer. Cnem. Soc., 1927, 49 435. 191.M. Mori and R. Tsuchiya, Bull. Chem. Soc. Japan, 1960, 841. 192.A. Haim and H. Taube, Inorg. Chem., 1963, 2, 1199. 193."Stability Constants", Chem. Soc. Special Publ. No.17, London, 1964, pp. 150 and 151. 194.R.G. Yalman, Inorg. Chem., 1962, 1, 16. 195.T. Dickinson, R.C. Irwin and W.F.K. Wynne-Jones, J. Electroanal. Chem., 1964, Z, 297.• 196.A.J. de Bethune and N.S. Loud in "The Encyclopedia of Electro- chemistry" , C.A. Hampel, Ed., Reinhold, New York, 1964, p.415.