The anodic oxidation of aqueous solutions of acetate ions at smooth platinum electrodes

JOL?RS.lL

532

THE XT

ASODIC

OSIDATIOS

SJIOOTH

PART

II

THE

/Grea#

of

AQUEOUS

ST_4TE

J_ R. AIXXSFIELD

Physical

ELECTROASALYTICAL

SOLUTIOSS

OF

CHEMISTRT

_ACETXTE

IONS

ELECTRODES

SOS-STE-4DY

aI_ FLEISCHM_X!SN, Department

OF

PIATINU~~

OF

Che33ristry,

OF

THE

_XSD LORD

Lr3literszty

of

KOLI3E

S-X-XTHESIS

OF

ETHASE

\VYSXE-JOW3

_YeulcastIe

rrfio3z

-Ty3ze,

_Xcmcastb

zrpmr

Tyne

r

Brifai3r)

(Received

July

Dedicated

to Academician

Ixth,

rgGg) A_ X_ FRCMKIS

on his 70th birthday

Monobasic carbosylic acid anions, RCO2-, number of products depending on the conditionsl. alkanes and Rz, formed in the Kolbe reactions,

alcohol ROH,

reaction 3_ In the steady-state

formed in the Hofer-Moest

of acid aqueous acetate solutions at smooth platinum dimer, ethane. The mechanis’m of this reaction may CH3 - COOfollowed

by either

2

or by dimerisation

- COO(M)

2

(I) after dimerisation

z CH3 - COO(M) CH3 - COO-tCH3

%

(CH3

2

(CH3 - COO),(M)

CH3 - CH3;z

- COO),(M)

(24 -I- e

CO3

(3)

(1%) “‘!

CHs

(M)

+

CO1

(4)

2CH3(M)%%H.-CH3 -(&I> +

The reaction being !&i&r

(2b)

after cbsproportionation CH3 - COO

Cl&

electrolysis

electrodes the main product is the be formulated as a discharge step

CH3 - COO(M)+e

disproportionation

(CH3 - COO)-)(M)

may be oxidised anod.icallJ- to a The products include the dimer, olefines derived from R, and the

CH3 - COO-

+

(5a)

CH3 - CH3

+

CO3 +

e

(zb)

path (I), (3) is related to the diacyl peroxide theory, the mechanism to that proposed by FICHTER~ e_ucept that in that theory an intervening

dticarb?xylation led to step (+). The path (I), (4), (5a) is the discharged ion theory5. Tlie intervention df hydrogen peroxide in the formation of acetate radicals has also been-postiated6. _ -_ In -the‘ reaction schemes outlined above, the electrochemical steps (zb). (5b) have- been included:-forcompleteness Land. by analogy to the hydrogen evolution reaction_ _ It may benoted~that~all the-possible reaction routes are kinetically similar. :

~~=&‘Gt,OtZ&_

.-

Chctiia,

ro.(rg65)

522+7

_. .

OSTDXTIOX

OF CHaCOO-

AT SSIIOOTH Pt ELECTRODES.

II

523

if the decarbox;lation steps are assumed to be fast, in that they involve step followed by either a second-order or a pseudo-first order step. In

common

information

on

with

the

studies

Kolbe

of

other

reaction

has

gas

evolution

processes,

been

obtained

from

the

a first-order bulk

of the

measurements

in

the

steady state or else by a “polarographic” techniquel.7. These experiments give data about the slow stage of the reaction_ It has been shown by GERISCHER AND MEHL~ that further information on the react&n mechanism of hydrogen evolution may be derived from the form of the current-time curve obtained on applying a step in potential to the workin g electrode, since the shape is dependent on the variation with time of the surface coverage bJ- adsorbed intermediates. Recently, there have also been applications of the method of derivin, = the concentration of adsorbed intermediates from the overpotential decay curve as proposed by ARMSTROSG _IXD BVTLER~. It has been stated that for the anodic decomposition of HCOOin HCOOH. which is related to the Kolbe reaction, adsorbed formate radicals follow a Temkin isotherm and that a decarboxylation step similar to (4) is rate-determininglo. For the coupling of CF&OOin CFaCOOH, the decarboxuylation step (1) or the electrochemical decarbox>-lation-desorption step (zb) were regarded as rate-deterrniningrr. The Kolbe reaction has also been studied by applyin g variable frequency alternating current to the electrode and it was again claimed that the first-order decarboxylation of the carbosylate radical on the surface is rate-controllingi”_ In this paper, results are presented for the coupling of acetate ions in aqueous solutions to form ethane, as studied by a new repetitive potentiostatic pulse methodra_ As

a number

of

Faradaic

processes

take

place

that

involve

the

oxidation

of

the

acetate ion, and as the electrode is also subject to oxide film formation, it will be shown that no quantitative data can be derived from the form of the current-time plots. On the other hand, the yield of ethane, and of other products derived from the acetate ions, may be measured as a function of the length of the oxidation pulse, the height of which determines the magnitude of the rate constants, and the duration, the concentration of the adsorbed intermediates. In this way the rates of individual steps rather than of the overall reaction may be followed and kinetic information may be obtained from the relaxation times of the processes. As the amount of product from an individual used.

pulse

is too

small

for

analysis,

a train

of repetitive

pulses

has

been

THEORY The nature of the information which may be obtained the non-steady state as compared to the steady-state yield, means of a simple example for the reaction scheme

ilk?B-(M) B-(M) Assuming

Rg

2

from may

measurements be illustrated

+ e

c

“Langmuir”

of by

(7) conditions,

i.e.,

rate constants

independent

of coverage,

= keC&-O)-kks’ibkdJ J_ EZecboanaE.

Chem.,

IO

(1965)

511-537

31.FLEISCHMAXX,

524 where $2 is the saturation w&‘obtain

coverage

3.

R.

(moles cm-“)

XANSFIELD,

W.

of the surface.

F_ K.

After

WYNXE-3OXES

solving

for f3

ksC_&i-ks'+-k7

The integral Yield of .prodtict C is given by _,

u.siq$ tfiy @ct. th& %7oC= 0. Defining the relaxation I

:,

-= .z.

.’

time t

ks~ai+ke%-& :a.,

Z is de&&nGn ed by the largest rate constant_ The criteria of reaction &khani~ti I&&~ be. built up bearing this in mind_ On the other hand, as is well -Imo>ti, ‘she &&dy -s&.&e

: si.ndiy&:~c& t&it

OSZD_ATIOX

The

CHzCOO-

OF

special

case

of the

AT

SMOOTH

Pt

ELECTRODES-

discharge-bi-radical

11

mechanism,

525

(kzb=o), which

will

be

of

particular interest, gives z n sinh “/ b cash y-l-n sinh y

O=

(1-6)

with b2=&-+2n,

2n

The current-time

k1 C-t-jys

=

transient

and the integral

yield

YtC=;I’

is therefore

1

(b -n)eY+(b+a)e--7 (b+a)eY f(b_-)e-3’

af b-4

bks,t a

Y--

of product,

which

(17)

will be examined

below,

ks,tPdt 0

I

= k 5n

-

( (a

(I-EQ?)”

I

-

;)

(a -

time

may

with time, a time,

(

and

if kICA-

Alternatively,

YF

ksn >

4 ksaB

A?’

the

of the differential

from ye1

(x6),

in view

equation

of the rapid

no simple change

of 8

and hence

s ks3

(I91

IT

discharge

step is slow,

If )

at short times,

=

kaak1”C%-

compared to kKAcontains_a

$- krC_c

ksakxCii(

However,

z, may be defined by

-g ksa i.e.,

I

-N z-

(I-F)]

(18)

of the non-linearity

be derived.

kz”CA-”

I

-= t

2(1 -p7) +ln

-

In (atb&)

where go= coth #_ Because relaxation

-

(a + %4 2

4a+&P) -

(=+q)]

2

(I-+~ 2 [

(&”

z(I+p)+ln

+

I

-b)?-

product

t2

(21)

3 a2

in the steady state. of rate constants

It can be seen that the characteristic time now but that these appear in a different combination

than in the steady state. In particular, than that of the steady-state current.

the potential J_

dependence

EZectroanaL

of z will be smaller

Chews.,

IO

(x965)

522-537

EXPERIMESTAL

The materials used in this study were prepared as described in a previous paperl3. The cell used is shown in Fig. I ; it was basicall_y a two-necked 25a-ml roundbottomed flask of Pyrex-glass ; the central neck terminated in an extended BgA-cone. A side-arm from this neck was the connection to the vacuum line. The second neck, directly opposite the side-arm, was used for carq-in g oblicquely, the reference electrode

Fig. line;

I. Cell for the collection (b). palladium-hydrogen

of the gaseous products of electrolJ-sis: (a), reference electrode; (c), subsidiary electrode:

attachment (d), working

to vacuum electrode.

and its compartment. This arrangement allowed an easy alignment of the working electrode with respect to the Luggin capillary. A Teflon-coated magnetic stirrer was included in the cell to speed the de-gassing, thus eliminating the necessity of tapping or shaking the cell. The use of the extended cones prevented the flow of the grease from the joints into the flask. The cathode was a platinum wire spiral with its central axis in the horizontal plane, and the anode was # s_w.g. platinum wire of r-cm length terminating in a small glass bead to ensure radial symmetry with respect to the cathode. An activated palladium-hydrogen reference electrode charged to the LY+? phase transition was used in the same solution as in the main compartment. This halfcell behaves as a hydrogen electrode at a reduced partial-pressure of hydrogen. Auxiliary tests were carried out to show that it was stable under the experimental (vacuum) conditions and that molecular hydrogen did not reach the main compartment_ All The potentials are referred to the normal hydrogen scale. The cell and the electrodes were prepared as previously described. The potento the working electrode using a potentiostat in tial-time progr‘amm e was applied -conjunction ivith-a pulse generator_ The electrolysis was carried out in the absence of air, i.e., at the vapour pressure of the electrolyte solution. In order to remove the air and residual ga2e.s from-the electrolyte, the cell was connected repeatedly to a conven@onal gas handling high vacuum nne having a working pressure of 10-6 cm Hg. The products bf _electrolysis -were removed from the cell using the same arrangement u&l successive periods of evacuation-gave no further yield of gas. After water and acetic acid fianbeen.removed with a-suitable arrangement of-traps and coolants, the gasesweie tr&sf&red to a gaS burette u&ng a sin&e-stage diffusion pump. The total

OSIDXTIOX

volume

OF

was

CHZCOO-

measured

AT

and

SMOOTH

the

Pt

gas then

ELECTRODES.

analysed

11

527

on a Metropolitan

Vickers

M.S.2.

mass spectrometer for mass numbers 2-62. This system was able to measure and analyse gas samples down to 10-3 ml in total volume, the average gas sample being 10-1 ml. In \-iew of the experimental arrangement and relative sizes of the working and subsichary electrodes, this electrode always remained in the vicinity of the h_vdrogen potential, and the anaiysis included the jet amount of hydrogen evolved at the cathode. This could be used under some conditions as a coulometric estimate of the net anodic Faradaic processes (see belo\\-) _ To establish the experimental conditions that prior to each pulse the surface experiments were coverage was zero, i.e., when f= o 0 = o, a series of preliminary carried out. As a result of these, the optimum base potential was found to be150 mV, this being the potential at which a maximum amount of reducible material was removed from the electrode after any particular oxidation potential and duration of oxidation time. The time interval between successive pulses requrred to establish this condition, was examined by noting the variation in the quantity of electricity used in the osidation and reduction. It was found that it was practical to have a dutv ratio of one in ten provided the interval exceeded I 5 x 10-3 sec. In practice the lolver Iimit used for the interval between pulses was 5 x 10-3 sec. The other parameter, the magnitude of the oxidation potential, was \-aried as required. The major part of the esperimental work was carried out using an electrolyte of aqueous I M sodium acetate-r 41 acetic acid, and for comparison, I flf sodrum acetateI i'k1acetic acid. The electrolysis was carried out for a series of oxidation oxidation potentials, FioX, for these solutions, times, to=, for four, and two, different 17rd. of -150 respectively, and a base potential, mV. The gas samples were analy,sed for hydrogen, oxygen, carbon dioside, methane, ethylene, ethane, methyl alcohol, formaldehyde and dimethyl peroxide (details of the procedure are available on applicationi’). The formaldehyde content of the electrolyte was determined by a method described by JOHKSON AND SCHOLES~~. It is based on the formation of the 2,4dimtrophenyl hydrazone and its quantitative extraction with carbon tetrachloride; the strong red colour, produced on th- addition of alcoholic sodium hydroxide to the extract was estimated photometrically--. Only very small traces of aldehyde were formed in the solution (barely above the blank determination)_ The concentration of methyl alcohol in the electrolyte was determined by a new methodle. This was based upon the complete oxidation of the alcohol in aqueous solution to carbon dioxide using a strong acid solution of potassium permanganate at an elevated temperature. Mna+ ions were removed from the sample as the die_xide, and the change in the permanganate concentration estimated photometrically by the change in amplitude of the 5,300-A absorption peak of permanganate. The method will easily detect 0.1 p-p-m_, but the amount of Lalcohol remaining in the solution was always well below this figure. EXPERIMENTAL

RESULTS XSD

Form of thecu7rewt-time The form can be seen that

IKTERPRETATIOK

cumes

of the experimental current-time curves is illustrated in Fig_ 2. It the shape depends on the base potential i.e., on whether the surface J_ EZectroanaL

CWem_.

IO (1965)

51-z-537

M. FLEISCHMANET,

528

J_ R.

Mz;\KSFIELD,

XV. P.

Ki. \VXTSE-JOSES

was initially free of oxide, or covered by the oxide formed at the higher porential. At long times, the transients approach a common constant current indeFendcnt of the history of the electrode. Comparison with eqn. (15) shows that falling currenttime curves are consistent with either a discharge-bi-radical mechanism or a dxchargeslow. For the first mechanism, eIectrochemica1 mechanism, the second step being the curves in Fig. 2 may be compared with eqn. (17) plotted in Fig. 3_ It can be seen that a rough fit may be obtained, but that the data indicate a fall in 8 with increasing overpotential, which is clearly impossible. Alsc, the curves in Fig. 3 do not fit eqn. (9) for the discharge-electrochemical mechanism which would demand a marked fall in

cm-=

A

,;_;I 10-4 Fig.

2.

NaAc-r

\-__>+.o” 10-2

1O-3

Experimental M HAG. (A),

I?ig.‘ ~z_-A redna a;~*.-

current-time base potential,

&&able

@lolot of ifie

10-l

transients - 150 mV;

cal&la@d

1.0

set

(constant (B), base

curkht=timi:

overpotential) potential, f

curves

for an 1,260 mV

for discharged

electrolyte

ibn-b&radical

I

nr

OXID.~TIO~

OF

CHKOO-

AT

SMOOTH

Pt

ELECTRODES.

529

11

inthe half-life with increasing potential. It is clear, therefore, that quantitative formation cannot be obtained from the plots. The analysis for products given below shows that in the region where the ethane yield is increasing with time, an alternative parallel Faradaic path, the complete oxidation of acetate ions, is diminishing. In this time region, the electrode is also progressively oxidised so that quantitative evaluation of the transients is not possible_ This is also implied by the break in the current-time plots in Fig. z lvhich shows that several Faradaic processes are taking place.

Yields

of prodzicts

as a fisnction

Hydroge~z n&

of time, potentid

and PN

carboiz dioxide. The integral yield of h3idro~en~cm~/sec as a

function of the polarisation time for various o,xidation potentials is given in Fig. 4 (a and b). It can be seen that the overall rate of the electrode reactions increases with

70

10-l ‘; it N

-/ ;rp

E

k

,o-’

-

\g

cl

2 lo;;+

1; p”ga-----”

I

10-3

O

v

0

1O-2

lOSl?C

10-*

(cl

Fig_ + The Integral X&C--~

( of,

z%$H_%c;

•i- z-55

yield

(b),

v;

I

( Af. +

of hydrogen

AfhTa_4c-0_1

z-75

as a function 32

I~//(

v t7

-

-

----___

----

17

,

2

10-2

-

H_lc_ Oxidation

10-a

of oxidation

potentials:

(b)

0 I

I

10-2

1osec

10-l

time for an eleCtrOlYte (V ). f

2-z vv”; ( 0).

f

(a). I fif 2-35

V;

v.

pH for the same oxidation

potential and time. The yiekls of carbon dioxide were found to be somewhat variable, particularly at the lowest oxidation potential. This was undoubtedly due to the difficulty of removing the gas completely from the soIution (which had a relatively high pH) and transferring the gas in the vacuum line. A convenient measure of the amount of carbon dioxide evolved is as a plot of the ratio to the amount of hydrogen against oxidation time for particular values of V ox. Fig. 5 (a and b). The pattern of behaviour may be more easily discerned from this plot than from that of the amounts of the single components, since any iraccuracies due to the variations in electrode area from one experiment to another are eliminated. Some uncertainty is introduced at short times where the yield of hydrogen is reduced, as an appreciable fraction remains adsorbed on the large subsidiary electrode. Bearing in mind the irreproducibility in the yield of carbon dioxide, the results show that at times less than 10-3 set the ratio is of order two, while at longer times it approaches one-half_ This can be explained if one regards two limits for the oxidation of the acetate ion. The lowest oxidation state is the discharged acetate ion CH3COO-

+

CH3COO-(M)

and disproportiohation

+e

of the acetoxy

CHaCO_O- (M) -+ CHs-(M) -I-COs

(24 radical gives one molecule

of carbon dioxide :

(23) x0 (It%%)5==--537

>I.

530

FLEISCHJIXNS,

J_ R.

1\I_AXSFIELD,

\v’. r;.

R.

\\‘SSXE-

JOSES

2

‘-b/C02

-f.O 0.5

4

I

I

I

lo-

lo-

10-3

*

I

13-l

i

0

Set

V

0.11

lo-"

I

10-X

1

-

I

10-2

10-l

I

l.Osec

Fig.-51 A plot of the ratio of the yield of hydrogen to carbon diosidc as a function of oxidation I +I NaAc-o-1 M HAc. O_xidation time ftir. an electrolyte (a), I .M NaAc--I M HAc; (b), pot&tie: (V ). + 2-z V; ( 0). + Z-35 V; ( O)_ i Z-55 V; ( 0). + 2 75 xv-

& the hydrogen evolved is a measure of the net Faradaic process, a ratio Hz/CO2 = 0.5 is found at-long times where ethane is formed from the methyl radicals. The complete o&I&ion of the acetate ion would yield two molecules of carbon dioxide --C+-(M)+

+C&+3H++3e

(24)

@in~wi~(&),-H&OS= 2. It is cleartherefore that complete oxidation takes place at &+I~ tin&L The-complete oxidation of acetate ions is more potential-dependent than theforinat&-r‘bf ethane. Et&&ze. The integral yield of ethane as a function of L.., given in Fig. 6a, will .f@st be d&&sed q&litativ&y. It is observed that the ste_ady -state for the three $igh+=pote&iSl& is &tt-airied. within IOWA-10-2 set of polarisation. and that the rel&ation----, -.:_ $1&e, change :in~ m&ha&m -e_tj&e~ foniration %I the potential region where the ~&&&$~er%&n lo2y&n fb kthane evolution. takes place in the steady state. The .f&~_&&e’&uGe$n Fi&6a agrees-with that of-the plots of H
of

OF CHKOO-

OXIDr\TION

of ethane

_4T SMOOTH

is a measure

of the

change

Pt

ELECTRODES.

in the

mode

11

of oxidation

531 oi the

acetate

ions.

Alternatively, the region in the vicinity of IO-~ set may be regarded as marking the onset of the inhibition of the osidation of the methyl group or, in a general sense, of hydrocarbon oxidation _ The results for I M sodium acetate-o.1 M acetic acid show a similar relationship between the yield of ethane and carbon dioxide and o_xidation time, in that no ethane was evolved; a detailed comparison cannot be given because of the limited extent 1

+ * !ff of our data

o-’

0 0

lO-2

0 0

olScr

!L

1 II

9 ”

(4

L

I”

-.

u 1o-3 z i

8

0

10-5;

:

r

v

0 II :

1o-41o-4 1o-2x5

I

10-Y

10-q 10-Z

I

I

10-l

10 set

t 40

0

\ 10-x

10-z

10-l

1.0

1

10

102

set

I .iLl Fig. 6. The integral yield of ethane as a function of oxidation rime for an electrolyte I &I HHAc. Osidation potentials of: (V ). + 2.2 V; ( 0). +- 2-35 V; (@), I 2-55 V; ( A). + (a), short total polarisation times; (b), long total polarisation times.

NZAC2-75

v.

In general, the total oxidation time used in the above experiments was 30-60 sec. An earlier series of results was obtained for total oxidation times of 5-30 rnin the sample being collected at atmospheric pressure and subsequently transferred to a vacuum line_ The results for the ethane analysis, only, are given in Fig. 6b. The pattern of these results differs from that in Fig. 6a in that there is a maximnm in the yield of ethane in the region of 10-1 set followed by a minimum to give a steady-state value at ‘approximately IO set: The for-m of the curves appears to be independent of overpotentiaL Since the only difference between the experiments is the total oxidation J_ EZec~roanaZ.

them..

IO

(1965)

522-537

AI_ FLEISCHJIAXN,

532

J_

R.

3IAKSTIELD.

W.

I-‘.

K.

LX-I-XXE-JONES

time, it appears that the electrode history is the critical factor_ It is likely

that the change in o_xidation state of the electrode affects the measurements carried out over long perio&_ The yields at short times must also have been inaccurate in the series

(Fig. 6b) because of the method of gas collection_ The quantitative evaluation of the results will therefore be based on the data in Fig 6a. 2-k wtilzor oxidation $rodrtcts. The minor products were: oxygen, methane, ethylene, methyl alcohol, dimethyl peroxide and formaldehyde. These are expressed as the percentage composition of the gas sample in Fig 7, excluding carbon dioxide.

8

t

a

0

lo-*

10-X

10-z

1o-1

1.0 set

Fig. 7_ The minor products of electiolysisExcluding the contribution of CO1 to the the_ 0/0 contributions are given as a function of oxidation time. For an electrolyte I A4 Hkc for potentials: (a), + s-35 V; (b). + 2.55 V;.(c), -i- 2-75 V- ( 0), oxygen; (a). ethylene; ( A). methyl alcoholperoside; ( 0). methake; (0).

gas sample KaAc--r M dimethyl-

A sigriificant feature is that the yield of oxygen is very small at times of IO-~ set but
OS1 DATIOW

Gr

CHnCOO-

XT

SMOOTH

Pt

ELECTRODES_

11

oxygen demonstrates that the oxygen evolution has 10-d-10-3 set a--d that this reaction must be associated Since the electrode potential is above that for oxygen

533

a relaxation time in the range with a change in the electrode. evolution and acetate ions are

present, the inhibition of the o e-r. with increasing time is not unexpected. The yield of methyl aicohol indicates that after the establishment of the conditions for ethane production, the bi-radical recombination is not the only pathway for removing methyl radicals_ The presence is further confirmed methyl

radical

of methyl radicals (possibly adsorbed at the electrode surface) by the vield of methane. As has been noted by other workers, a

is able

to

abstract

an a-hydrogen

from

an

aliphatic

acid

to

form

methanei’. CHB+CH&OIH The radical that dibasic acid I

-> CHJ+CHH~COZH

is produced CHz.

zCH~~-CO-~H

+

as a result

(25)

of this reaction

is able

to dimerise

to form

COaH

a

(26)

i

CHz - COeH

The dibasic acid under the conditions of the electrolysis wilI then decarboxylate to yield ethylene, and, since the concentration of CHzCOlH will not be high, the frequency of dimerisation will be low thus accounting for the low yield of ethylene (compared to methane). Other products must result from the hydrogen abstraction reaction, but they do not appear in measurable quantities. Finally, the presence of methyl radicals is also shown by the formation of dimethyl peroxide and it is clear that acetoxy radicals decarboxylate to methyl radicals as an intermediate step in the o_xidation. The relaxation time for this process is also clearly smaller than that for ethane formationThe presence of the particular minor products may thus be seen to be consistent with other analytical data and, m addition, the range of products is internally self-consistent_ DISCUSSIOE;

Measurements on the anodic oxidation of carboxylate anions have shown that in the steady state the yield of oxygen diminishes and that of the Kolbe dimer increases as a transition region at + 2.1 V7.le is traversed. For the non-steady state, above the transition region, this statement may now be augmented: there is a transition with increasing time such that the complete oxidation of the anion changes to give the Kolbe coupling. At short times, above the transition region, the evolution of oxygen is completely inhibited whereas carbon dioxide is formed from the anion. It therefore appears that the acetoxy and methyl radicals, formed in the discharge, compete for an intermediate in the oxygen evolution reaction, a stage succeeding the discharge step in this process being slow. The “uncontrolled” oxidation of the acetate anion at short times may be compared to the similar “uncontrolled” oxidation of other

.

compounds on platinum at lower potentialsis. With increasing time, the yield of oxygen goes through a maximum dimethyl peroxide (undoubtedly formed by the coupling of two methyl radicals /_ Ekchoanal.

Chem.,

IO (1965)

and with

522-537

31

534

FLI_;ISCHAI_\SX,

J_

iTi. M.\S>TII:LD,

W.

r.

K_

\VYSSE-JOSES

the osygen di-radical) follo~vs a similar pattern. T!r~-1-.epror:c *\ces are clearly a.;sociated with special sites and occur transientllas the electrode is co\-t’rr:d by an oxi& la\-t-r. We suggest that the reactions take place at the periplrcr>- of t~vc~-dill~en~ional ccn;rrs of the oxide groGng on the platinum substrate; the cstcnt vi tlris pc:ripl~~r~ pabs~s effects Louvrealso been through a maximum with increasing time~o_ Similar catalytic observed

for hvdrogen

evolution”1.

On the basis

of this interpretation,

the

complete

oxidation of the acetate ions takes place on the platinum substr-ate xl-bile the Kolbe coupling is restricted to the surface covered bJ- the oside layer. The role of the osrde layer (probably in part controlled by specific cation adsorption’“) in determining coupling reactions and, indeed, of oxygen evolution, has also been recognised from steady state studies. The measurements in the non-steady state, howelrer, suggest that the formation of the oxide layer is the primary cause for the inhibition of oxygen evolution whereas other workers have attributed it to the adsorption of anions. Reference to the non-steady state of oxidation of the surface”3 shows that the layer is not in its final form m the time region in which the ethane evolution reaches a steady state. This may well explain the much higher rate of ethane production, Fig. 6a, than is observed in measurements carried out over long times (differing by at least one order of magnitude) The formation of methyl radicals before the formation of ethane in the present measurements, is also supported by other experiments such as the formation of trinitro-xylene from trinitrotoluene in the presence of sodium acetate’4 as well as the Relatively stable alkyl radicals are also methylation of butadiene and isoprenez5. implied by the formation of inactive products from s-methyl butJ-ric acid”=. It is likely that the alkyl species are adsorbed (see below) while the formation of carbonium ions in the present reaction is unIikeIy”7. The ready formation of alkyl from carboxy radicals is at variance with rate-determining steps that have been postulated The evidence for the formation of a film of carboxy for the Kolbe reactionlo-1’. radicals obtained from discharge and decay curves 10.2s on which a rate-determining decarboxylation was based. cannot be regarded as unambiguous, since under oxidising conditions a layer of surface oxide may weIl be formed even in non-aqueous solutions*. Such Iayers are known to react with organic compounds during the decay processag. Jn fact, in-the oxidation of acetic acid as measured in the present investigation, the initial state 8=0 is probably formed at the base potential by the oxidation of methyl radicals on the- uncovered platinum surface as the oxide layer is reduced_ The ready formation of methyl radicals during the reaction might be thought to-cast doubt on the kinetic formulation of eqn. (14). as it is, strictly speaking, necessary to divide the reaction schemes into two parts, one referring to the formation -of acetate -_Ith coverage 0 -(for example steps (I) and (4)) and the other to the formation of methyl radicals with coverage 19’ (e.g. steps (4) and (5)) ;

r

.?

-

Arate-deternzinin g.dischargewasalsoruled TafeIsIopeobserved experimentout loasthehigh ally could-notbederived~tider"Teinkin" corditi ons.Ihis result,however,depcndsona.?6unCng r&action (1)t.o beinn~uili&-iuti. Iti3 &adily &onfii?medthatfor the mechanism g&ing eqn. (14). high-Tafelslo~_canalsoapplyfara-raterdetermining discharge ;nder"TeITlkin"conditions. -A

I I_ EL+~oa_tiuJ_

CW_i

IO

(ig65,)

5z?-=p~7

OSIDATIOS

OF

CHXOO-

AT

SMOOTH

Pt

ELECTRODES_

11

535

The simplest case arises if ksa > ka > klC_a. It can then be readilv seen that the formation of ethane shows two separate relaxation times, and the consecutive reactions become effectivel\uncoupled The relaxation time of the larger amplitude step on the yield-time curl-e is then no longer controlled by the fastest decarboxJ-lation reaction

and the reaction

scheme

reduces

to eqn.

(14).

It is to be noted

that

only

one relaxation is in fact observed esperimentall>-. The pattern of the results for the ethane evolution, Fig. 6a, can be seen to fit the discharge-bi-radical mechanism in most respects, the discharge step being slow. The slope of the logarithmic y-ield-time plots is 1-7 which approaches the value a required by eqn. (21). In addition, the relaxation time decreases only slightly with increzing potential and it can be seen that such a small variation is demanded by eqn. (20) for this mechanism. Too much emphasis cannot, how-ever, be placed on this observation as the steadv-state yield as given by Fig. 6a itself shows a much smaller potential-dependence than the data obtained from experiments using long polarisation times, Fig. 6b. \Vith this assignment of mechanism, the rate constants RI and ksa may be derived from the steady state and eqn. (20) or (21). Equation (21) is to be preferred, since the level of approsimation is better defined. The results are tabulated in Table I.

+2_75

3-I

x

10-J

0.0 x

10-3

4.7

x

10-i

+-'_55

1.5

x

10-J

5-G

10-9

2.5

%

10-T

10-J

1.9x

10-3

I.5

_x

10-T

+

Z-35

Using

1.3 x CA-

=

xo-J

moles

crr~--~, 0

=

10-0

s

molts

cm-2

(with

a reaction

zone

thickness

b N

IO-?

cm)-

The fact that the constant ksa b/i2 or rate ks,, derived in this way is found to be smaller than the constant k1 or rate klCA may be thought to be at variance with the assignment_ Several factors however combine to make the estimate of ksa only a lower limit for the actual value. In the first place, the rate constant kl refers to the apparent area and the correction for any roughness factor will decrease k1 and increase ksn as derived from eqn. (20) or (21). Secondly, the fact that ethane is only formed on top of the o-xide layer implies that the relaxation time may be displaced along the time axis by the delay in film formation and this would also lead to an increase in the true value of ksa. The most serious effect would arise if the Kolbe reaction were in steady state equilibrium on the surface of the oxide patches, i.e., d0/&=0 in eqn. (r4), since the kinetics of film formation’0 would also give a yield proportional to the square of time as in Fig. 6! The degree of coverage of the electrode by oAxide”3 however- argues against this possibility. Thirdly, the rate constants for decarbo_uylation (4) and dimerisation (5) may not be sufficiently different to permit an assignment of the experimental relaxation time to the dimerisation step only (see argument concerning eqns. (27), (zS) above), an effect which is similar to the relaxit may not be valid to treat the nonation of coupled reactions in solution aO_Finally, steady state assuming “Langmuir” conditions_ lMaking the simplest possible alternative assumption of “Ternkin” conditions31, eqn. (14) must be replaced by J_

~~ectroana~.

Chent..

10

(rg65)

522-537

31. FLEISCHXXSS,

536

J_ R.

WV. 1:. K.

X-‘.SSFIELD,

\Vl-SSE-JOSES

(29)

where LUand p refer to the effect of the change in the heat of adsorption with co\-erage on the energies of acti\-ation of reactions (I) and (sa). Making the usual assumption that

the

terms

coverage,

before

the

the steady-state

ymc

K

exponential yield

do not

is determined

vary

rapidly

in the

middle

range

(F)s’(a-s)

(30)

On the other hand, in the initial stages of the reaction dependence of the coverage by the linear equation 8 = klCA-

we ma>- approximate

the time-

t

and if t is re-defined obtain

of

by

(31) as the time

at which

the yield

reaches

a .s~wzZlfraction

of Ymc

we

(32)

[X3 ,4 more accurate expression can, in principle, be obtained by replacing (31) by a Taylor series expansion, if necessary as a function of the parameter ks,/klC_l-. Equations (30) and (32) may be contrasted with a steady-state yield of klC_4- and eqns. (20) and (21) under “Langmuir” conditions_ We may anticipate that LYc p. since bc arises in the discharge of one ion whereas fi is connected with the bi-radical step. Defining a small deviation from Langmuir behaviour by CK= A p with A being small we obtain, Y&

oc &CA-) I

i+Y,C

cx

l-akd

,

(33)

(klc-d”“K5al_“”

fi22

so that the individual rate constants cannot be accurately determined from the steady state and relaxation time without a knowledge of d_ The effect will again appear to increase kl and decrease ksa. It is not possible at this stage to assess the relative importance of the various complicating factors. The balance of the evidence and, in particular, the very sharp relaxation with decreasing time appears to be consistent only with a dischargebi-radical mechanism. SUMMARY

The lc+etics bf the Kolbe electrosynthesis of ethane from acetate ions at smooth platinum electrodes in aqueous solutions h& been examined by a new repetitive potential pulse technique. The integral yields_ of the products of electroiysis have been measured as a function of oxidation time. At short times acetate ions are oxidised completely to carbon dioxide. With increasing time this process is inhibited and the alternative ox&ation to ethane incrkties to the steady state value_ at 10-3 sec.The yields bf these -and the minor products are related to the catalytic activity of the plati&m substrate. From the form of-the variation of the yield of ethane with time and it‘s dependence on $fential the discharge:bi-r5dical mechanism has been assigned

OSIDATIOS

OF

CH&OO-

to the Kolbe reaction, given

for the

AT

SMOOTH

intermediate,

of surface deviations

ELECTRODES_

537

11

the discharge step being rate determining.

heterogeneous

rate

effect of a number of complicating dependence

Pt

oxide from

constants

of the discharge

and

Limiting bi-radical

values are steps.

The

factors (the ratio of the real to apparent area, time

formation, “Langmuir”

the

influence

conditions)

of

more

are briefly

than

one

adsorbed

considered.

REFERESCES I G. E_ Svx~~~ovs~x~x XXD S. A. \-OITKEVICH. Rzlss. Chem. Rev. E?zgZirA Tvansl.. 3 (1960) 161. z H. I
~kCtYOaTZd_

Che?~l.,

IO

(1965)

522-53'7