The Mechanism of the Cathodic Hydrogen Evolution Reaction'

J. O'M. BOCKRIS AND E. C. POTTER

Imperial College of Science, London, England

ABSTRACT Some of the o u t s t a n d i n g problems of concept and m e c h a n i s m in the field of cathodic h y d r o g e n evolution kinetics are discussed a n d clarified. A full d e r i v a t i o n and correlation of kinetic e q u a t i o n s which assume no m e c h a n i s m reveals expressions for several p a r a n m t e r s which take wflues specific to one or more mechanisins. The use of statistical methods of t r e a t m e n t of d a t a proves indispensible in e s t i m a t i n g the values of these parameters. A f u r t h e r m e t h o d of ascertaining the m e c h a n i s m of the hydrogen evolution reaction is to examine the kinetics of the individual reaction paths, so t h a t the expected values of p a r a m e t e r s conmmn to all p a t h s m a y be deduced. A n u m b e r of mechanisms i m p o r t a n t in acid and alkaline solulions are thus t r e a t e d , and are shown to be dist i n g u i s h a b l e experimentally. Using already published data, the actual conditions under which various reaction p a t h s take place at mercury, silvcr, nickel, and smooth p l a t i n u m cathodes are calculated. I t is not only possible to compare these deduced d a t a with obs e r v a t i o n , a n d t h e r e b y verify the occurrence of a p a r t i c u l a r reaction p a t h , b u t also to d e n m n s t r a t e the impossibility of some m e c h a n i s m s in specific cases. The recent advances which the foregoing m e t h o d s have made possible are discussed in r e l a t i o n to d a t a which have r e c e n t l y beconm available.

I. INTRODUCTION The electrolytic evolution of hydrogen is generally chosen as the main subject of experimental and theoretical research in electrode kinetics because it was originally thought to be one of the simplest electrode reactions. This expectation has not been confirmed. However, work already done has emphasized the importance of the reaction, and its practical significance to studies in corrosion has stimulated further attempts to solve the problems discovered. In the following, some general equations of electrode kinetics are formulated and indicate the experimental methods best suited to mechanism determinations. II. SOME DIFFICULTIES OF TECHNIQUE AND CONCEPT It is notoriously difficult to measure the velocity of an electrode reaction and the corresponding electrode potential with reproducibility. The best agreement obtained between workers in different laboratories is + 8 my on Hg and =t=20 my on solid (e.g., Ni) cathodes. Many different ~ypes of Tafel lines are reported in the literature. Some of these are collected in Fig. 1, from which it is seen that only the first and the last two represent forms which are free from vitiating influences. The principal vitiating factor in experimental work arises from mimlte traces of impurities in solution. For example, addition of only
M a n u s c r i p t received M a r c h 26, 1951. This paper prepared for delivery before the W a s h i n g t o n Meeting, April 8 to 12, 1951. 169

10-1~ gram moles/liter of As203, CS~, CO, KCN, etc., detectably affect the electrode potential at a given current density at nickel cathodes (1). Reproducible results can only be obtained in solutions which have been purified by pre-electrolysis on to an auxiliary cathode (2). Use of this method involves laborious work to determine the amount of preelectrolysis necessary for a given electrode and solution. The criterion by which the optimum conditions for pre-electrolysis are found is that passage of further coulombs at higher potentials makes no further difference to the experimental results. Even then, the initial state of the solution before purification may be difficult to reproduce so that sometimes less pre-electrolysis is necessary than at other times. Many difficulties in preparing the surface of solid metals in a reproducible, clean state have caused most studies to t)e made on mercury. In one method (3) the electrode in wire form is heated in a stream of hydrogen to remove oxide films and sealed into a thin glass bulb containing pure hydrogen. After pre-electrolysis of the solution the bulb is broken by a glass probe thereby immersing the electrode in the purified and oxygen-free solution. A further possible method would be to heat the wire electrode electrically, e.g., inductively inside the cell, containing hydrogen, before placing it in solution. In addition, a few difficulties connected with the general attitude of much past work to studies of the hydrogen evolution reaction may be mentioned. First, the difficulty of measuring the electrode poten-

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

170

JOURNAL OF THE ELECTROCHEMICAL SOCIETY

April 1952

tial, and the comparative ease of measuring the current, directed attention away from the fact that the problems coflnected with overpotential were purely kinetic, and that attention should be focused upon the current. Second, considerable attention has been paid to the determination of i0, the current
DEPOLARIZED

difference between the metals. If AV is a potential difference measured by means of a potentiometer connected to wires of composition M, then, A V = VM-- V p t + X . Now, X ~ ~M -- qh,t, where q)M and ~p~ are the work functions of the metals M and Pt respectively (neglecting the Peltier heat in the usual way). Hence, VM = A V + V p t - - ~ M + ~ p t . Or, if the experimental conditions are appropriate, so that AV represents a hydrogen overpotential, VM = ~ + V r t - - r + r (1)

D I S S O L ~

O LIMITINGCllfiRENT ~ it ~1 R[$I$1A~IC-E ERROR i - T Y P E /

O ,~[ OXIOE FILM TYPE

~
i

where K is a constant potential difference for all cells in which hydrogen overpotential, 7, is measured at the same pH and temperature. But also, the relation between current density, ic, and potential, for a kinetic electrode process in whieh the reverse reaction can be neglected, is given (see below) by an equation of the form,

I ELECTROCH ICAL~. TYPE EM SILVER J

where a and K1 are constants and the pH is assumed to be constant. Thus,

LOG CURRENT DENSITY----~LOG CURRENT DENSITY

ic = Kx exp

[--a(~-

RT

"

(2)

FI~. 1. Some types of Tafel line for hydrogen deposition

16 14 t2 I0 ,_o
i

Therefore, if when ~ = 0, ic = i0, then,

I i i i i i t

[ = K~exPk| RT J'

(3)

6 4 2 0

MOFeW

3.0

. ,I

3.5

4.0

4.5 5,0 dp IN s

5.5

6.0

6.5

7.0

7,5

FIG. 2. Relationships between exchange current and thermionie work function for a number of cathodes.
passing in either direction at the electrode at the reversible potential, i.e., at zero overpotential. Consider a complete galvanic cell at constant temperature consisting of a metal M (the cathode in a cell in which overpotential is measured) and a Pt electrode in the same solution as the metal M. There are three potential differences in the cell: VM, that at the boundary M-solution; Vpt, at the boundary Pt-solution; and X~ the contact potential

where K2 is a new constant. A relation of type (3) is indeed observed, and is shown in Fig. 2. It follows, then, that the most appropriate potential at which to compare the rate of the hydrogen evolution reaction at various electrode materials is not the reversible hydrogen potential but the absolute zero of potential. In the absence of reliable knowledge of this, the potential of the charge-free surface appears to be the best reference potential. Similar conclusions apply to the heat of activation, AH*, which is usually quoted at the reversible hydrogen potential. However, measurements of i0 and AH* retain considerable relevance because they assist in distinguishing, between various possible mechanisms of the electrode reactions. Lastly, the rate of the evolution reaction depends upon the difference of two inner potentials $, i.e., it depends upon a Galvani potential, where 4~ is defined by ~b-- ~b+ x, (4)

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Vol. 99, No. ~

CATHODIC H Y D R O G E N EVOLUTION R E A C T I O N

171

~b being the Volta potential of a phase and x the surface potential. The latter potential is strongly dependent upon the surface condition of a phase, and hence the overpotential at a given rate of reaction would depend sharply upon trace impurities.
I I I . STATISTICAL METHODS IN MEASUREMENTS OF

In these circumstances the required variances are calculated from equations of the type, V(/5)
_

nr(nr-

E(pi-

p)2

1)'

(10)

HYDROGEN ELECTRODE KINETICS In order to observe with reasonable and known accuracy those of the theoretically important parameters in overpotential studies (e.g., temperature and concentration effects) which may vary little more than the reproducibility of the measurements, it is essential to replicate the observations and to apply statistical methods. In applying the method of least squares to estimate the parameters of a Tafel line, those formulae should be employed which correspond to the current as the independent variable and the potential as the dependent variable. This means that the sums of squares of the vertical deviations of the experimental points from the computed Tafel line are minimized, and that, consequently, a prediction of potential at a stated current has the minimum possible error based on the original observations. Thus we have
a = ~ b2,

where V(p) = variance of p, pi = the value of the parameter p of the -th j Tafel line,
p = (Zp~)/n,

and

nr = number of Tafel lines in the population. It can be shown that the relation between log i0 and absolute temperature, T, is log i0 = log B AH* 2.303RT' (11)

(5)

and
b - nZxy n~x ~ Zx2y _ N IF,x) 2 D '

(6)

where x and y take the corresponding experimental values of log i~ and ~, respectively, n is the number of pairs of observations, 9 is ( X y ) / n , and ~ is ( ~ x ) / n . The variances of the Tafel line parameters may be calculated from
V(b) = [n~Y2 - (~Y)2]D (n -- 2 ) D ~-

where log B contains a temperature term, but is usually considered to be temperature independent over the ranges of temperature often used experimentally. It is evident from (11) that since log B is formally the value of log i0 at infinite temperature, the value of log B is obtained by a long extrapolation of the observed relation. Also, since log i0 is itself obtained by lengthy extrapolation of a Tafel line, it is not surprising that published values of log B show lack of agreement (4). In order to obtain the most probable values of AH0* and log B from the data, it is clear that by treating 1 / T as the independent variable and log i0 as the dependent variable, equations analogous to (5) to (8) can be applied. The experimental accuracy which must be ac.hieved to attain any desired limits of error in an estimate of log B may be calculated in the following way. The variance of log i0, V(log i0), is given with sufficient accuracy by V(log i0) V(m)D' n' '

(12)

N2
' (7)

V i a ) = V(b) [ D / n + ~2],

(8)

and V(log i0) -- V ( b ) [ D / n + (log i0 - 2) 2] 9

be

(9)

where V(b), V i a ) , and V(log i0) are the respective variances of b, a, and log i0, each with n - 2 degrees of freedom. When replication of observations is carried out, it is often desired to replace the population of Tafel lines so obtained by one mean line. The parameters of this line may be calculated by bulking the observations and applying (5) and (6), but the variances of these parameters cannot be found by applying equations (7) to (9) to such bulked data.

where V ( m ) is the variance of m, m is H*/2.303R, D' is the appropriate denominator corresponding to D, in (6), and n' is the number of pairs of observations of log i0 and 1 / T . Also, the variance of log B, V(log B), is obtained from an equation analogous to (8) as, V(log B) = V ( m ) [ D ' / n ' + ~2], (13)

where ~ = Y r / n ' = ( F . 1 / T ) / n ' . From (12) and (13) and noting by analogy to (6) that D ' / n ' = Z(r - e)~,
V(log i0) =

V(log B ) Z ( r -- ~.)2
+ _

(14)

In order to use (14) ntlmerically suppose that the values of r are derived from the experimental

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

172

JOURNAL OF T H E E L E C T R O C H E M I C A L SOCIETY IV. GENERAL EQUATIONS

A. Current and Potential

April 1952

temperatures 0 ~ 10 ~ 20 ~ 30 ~ and 40~ and furthermore, let it be arbitrarily decided that the maximum inaccuracy tolerable in log B corresponds to 95 per cent confidence limits of 4-0.5. Hence, assuming that the errors in log B are normally distributed, we have 3.18 x/V-(log B) = 0.5, where 3.18 is the value of Student's t for 3 degrees of freedom and the 0.05 probability level. Hence from (14) V(log i0) is 0.000286. Now, from (6) and (9) it is seen, V(log io)
V(b)[(logio - log~)2 + Z(logi~ - logic) 2]
52

Suppose that ki is the specific rate constant of an unspecified rate-determining step in the forward direction of a reaction (ions depositing). Then,
kT ( k~ = K-h- exp

(AG*)I~ ~ /,

(16)

(15)

where log i~ = 2. This equation may be applied to typical experimental data for the mercury cathode, which (reference shows) gives the most reproducible overpotential measurements. Let log i~ take the nine values, - 7 , - 6 . 5 , - 6 , . - . - 3 , so that the value of log i~ is - 5 . 0 . Taking b and log i0 for the mercury cathode in aqueous acid solution as 0.115 volt and -12.0, respectively, and using the value for V(log i0) calculated above, it is found that V(b) is 0.0592 9 10-6 (corresponding to 95% confidence limits in b of 4-0.0006 volt). Such an accuracy in estimating the slope of the Tafel line is unattainable even with modern techniques. This is evident since an error of only 1 mv in estimating the overpotential at 10 7 amp/era 2 and a similar error in the opposite direction at 10 3 amp/era 2 is sufficient to cause an inaccuracy of 4-0.0005 volt in b. In a recent investigation using nickel cathodes (10) one twentieth of the experimental Tafel lines attained less than the above inaccuracy in slope, but the slope of the mean line of a population of Tafel lines could not be determined with an accuracy (95% confidence limits) greater than 4-0.004 volt. The 95 per cent confidence limits for log B were on the average 4-2.2, each limit being based on an average of 25 pairs of observations. It, therefore, appears that the quantity B is not a useful distinguishing criterion of reaction mechanism. While variances (which should be quoted with their number of degrees of freedom) are independent of distribution of errors they are not readily assimilable as measures of error, and it is customary to assume (in the absence of evidence otherwise) that the distribution of errors is normal so that confidence limits may be quoted. Some caution is necessary here; for if it is assumed that the parameters b and log i0 of the Tafel line have normal distribution of errors, then the errors in i0 itself are not normally distributed. Consequently, the distribution of errors of X which is calculable from i0 (see later) is not normal, although it is unlikely that serious error would arise by assuming normality.

where (AG*)~ is the standard free energy of the activated complex of the rate-determining step with respect to the initial state (here assumed to be X gram ions of hydrogen ions constituting part of the monolayer adjacent to the electrode surface, often termed the Helmholtz double layer) of the reaction, where ~ is the transmission coefficient, and where ]c and h have their usual meanings. Let a, be the activity of H30 + ions in the initial state. Then the forward velocity ~ is
: al]r

(17)

Also, the forward current is ; = ~XF, (18)

where X is the number of electrons necessary so that one act of the rate-determining step can occur. Suppose a potential difference AO~ is applied between the electrode and the initial state of the reaction. If this potential difference is positive it retards the flow of reactants over the energy barrier of the rate-determining step, i.e., it makes (AG*)I more positive. The potential difference increases (AG*)I by /~XFA~b~ where /~2~q~cis the potential difference through which the electrons pass before they reach the transition state. (Work done on the system after it has passed the transition state does not affect the velocity of the reaction.) Hence 0 < f~ < 1. varies with the reaction mechanism, and is a complex quantity, except in certain simple mechanisms. Thus, in the discharge reaction H~O + ~- e -+ M H -t- H20, ~ = 1 if the energy barrier is symmetrical. General and limiting values of ~ are evaluated below for common mechanisms. From (16), (17), and (18), = KXF -ff a~ exp By a similar argument, i -~ K},F aF exp (20) .( (zxa*h-(1~)x~+~F) ~7~
RT

(19)

where (AG*)2 is the standard free energy of activation of the reverse reaction referred to the initial

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

VoL 99, No. 4

CATHODIC H Y D R O G E N EVOLUTION R E A C T I O N

173

state of this reaction, and aF is the activity of the entities in this initial state. If 5 ~ = A~b,, the reversible potential, -- i ' = 0, where ~ = i0 = ~. (21)

By expanding exponentials, equation (24) becomes, i~ = Xi0~F RT (32)

From (19), (20), and (21) and since v = A~b~-- &b~, z = io exp \ - ~ ), (22) (23)

Thus, i~ and ~ are linearly related at sufficiently low overpotentials. Also, from (32) and formally allowing for the possibility of Oi~//O~?depending on 7, X = -- Fi0 \ ~ / , - . 0 " (33)

= io exp (-~ (1 -RT~)XvF~],

= io L e x p k , RT )

(24)

-- exp ((1 - ~ X v F ) I "

Equation (24) is the most general expression for the cathodic current, and represents a more fruitful form of the relationship than has been stated hitherto. Special Case 1. Relation between i~ and y at appreciable overpotentials (e.g., v more negative than about - 7 5 my, see special case 3). Equation (24) becomes i~= i0[exp(
or

X (see Ref. 21) can be termed the electron number of the reaction and is estimated experimentally by applying equation (33). As shown below, X is a valuable diagnostic criterion of reaction mechanisms. Special Case 3.--Relations connected with nonlinearity between ~ and log ic. From (24), i~ = i0exp where f = exp \ R T ] " (35) ~XnF~ [1 - f], RT] (34)

~XT/F~/j,

(25)

RT RT v = ~ In i0 -- ~,~, in i~. p^~ Equation (26) is Tafel's equation,

(26)

= a - b log io
or

(27)

When the departure of the Tafel line [see equation (27)] from linearity due to the reverse current in (20) is just detectable experimentally, then f is just distinguishable from zero, and v = ~. The smallest detectable and significant value of f, i.e., f.~, depends on the experimental design, but, provided a minimum number of about five pail's of observations fall within a region about 0.025 volt on either side of ~.~, it may be assumed that f~ is 0.05. Therefore, from (35), f~ = 0.05 = exp \ R T ] , or at 20~ (36)

RT = a -- a F In i~

(28)

so that, comparing (26) and (27) and (28), it follows that:

with ~ expressed numerically in volts, 0.075 X = -(37)

RT a = ~ In io,
b 2.303RT - - , ~hF a = fiX.

(2~) (30) (31) Equation (37) gives an alternative method of estimating X experimentally and has not been previously described.

B. Potential and Time 1. During Build-up of Overpotenlial

For a completely polarisable electrode, the net constant cathodic current is the sum of the condenser (i') and faradaic (i') currents:

Equation (31) shows that a is a composite quantity, and need not be between 0 and 1 since fl and X can have maximum values of 1 and 2, respectively. Thus, the experimentally observed values of a of 2, which have been difficult to explain hitherto, are a special case of equation (31). Special Case 2.--Relation between ir and v at low overpotentials (e.g., n less negative than - 2 0
mv).

i~ = i' + i'.

(38)

Let C be the differential capacity of the electrodesolution interface. Then,

- C d~t = i' dt

(39)

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

174

JOURNAL OF T H E E L E C T R O C H E M I C A L SOCIETY

April 1952

where dyt and dt are infinitesimal changes in m and time, t, respectively. Let ~t and % be values of at times t = t and t = or respectively. Since i" is given by (24), then from (24), (38), and (39), Cd~ = i~ - io exp at

Or, integrating between h and t2 and the corresponding values of A~,

h t~ -

[ (

CRT [e io~XF

RT-/In

x,.F 7

(An~ -

A~2).

(48)

-- exp ((i -R~XntF)].

(40)

Special Case 1. dC/d~ = 0; ~t more negative than - 7 5 my. Equation (40) becomes -Cd'-~t=i~-i~ Let i" _ fl~_~F) ] . (41)

Equation (48) indicates that during the initial build-up of overpotential, ~t relatively rapidly reaches practically constant values. For example, with 2.303RT/B~F = 0.1, the time required for overpotential to rise from within 10 mv to within 1 mv of ~ is the same as that required for the interval 1 my to 0.1 mv from %. 2. During Decay (i) Relations in which it is assumed that dC/d~ = O. From (41) with i~ = 0, and *t~ more negative than - 7 5 my,

io

f,

(42) dt - C exp , (49)

and suppose.f' < 0.05 so that i" is negligible in (38), then from (38) and (39), io = Cdm dt " (43)

where n't is the overpotential l seconds after the cornmencement of decay. Integrating, t -- io~XF exp \ R T ] -F eonst. At t = 0, 7t~ = ~ . Hence, Const. = -i0/3X~ exp From (50) and (51), t - i0~XF e x p \ R T ] exp . (52) . (51) (50)

Hence, from (42) and if f~ < 0.05, d~t/dt is a constant. To find the region of overpotential in which this result is valid, we apply (38) which becomes i~ = i" at t = ~ , so that from (25), i~ = i 0 [ e x p ( Also from (25), /~F)]. (44)

i" = io [exp (fl~F)].

Thus, from (42), (44), and (45), f'= exp ( f l X ( ~ % re)F).

(45)

Special Case 1. ~'t = % - A{, where Av' is less negative than - 2 0 my. Using this condition in (52) yields exp RT ] flXFi~t CRT

(46)

For example, let f ' = 0.03, then from (46), taking a typical value of 2.303RT/~XF from experiment as 0.12, % - w = -0.15, i.e., the relation of potential to time is linear [(43) is valid] to within 0.15 volt of the constant value n~. Special Case 2. dC/dy = 0; ~ more negative than - 7 5 mv; (v~ - vt) less negative than - 2 0
mg.

(53)

Having regard to the condition for A~', (53) becomes , iot exp . (54)

Let ~ - ~t = An; at t = h, A7 = A ~ I and at t = t~, A~? = AV:. Hence, from (40) and (44),

d,,_dt c.i~ ( /~F)]

(47)

9 [1-

exp

\--R-T/J"

Equation (54) shows that the initial decay of overpotential is linear with time. Hence, linear extrapolations to zero time in the commutator method of measuring overpotential are valid under the above defined conditions. Special Case 2. A~' more negative than approximately - 5 9 mv (normal decay); nt more negative than - 20 mv.

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Vol. 99, No. ~

CATHODIC H Y D R O G E N EVOLUTION REACTION

175

Applying this condition to (52), and thus neglecting exp(~X7~F)/RT gives,

t - flXFio
9 7t = ~ ".

exp ( xT;F]" \ RT
RT CRT

(55) (56)

RT

In t -- ~--~ In BXFi0--"

Equation (61) enables estimates of differential capacity obtained from the build-up of overpotential to be corrected for faradaic current from observations of the corresponding decay curve (22). The differential capacity may be obtained from the decay curve by application of (63), it being also necessary to know the values of b and log i0 of the corresponding Tafel line.

Hence from (56) and (30), dTi d(log t)

-

2.303RT /~XF

C. Current and Temperature 2

b. (57)

From (19) and (21), t~T ( io = ~XF ~ - al exp (AG*)~

-

Equation (57) shows that overpotential decays logarithmically with time under the above conditions and that the slope of the logarithmic decay curve is the same as that of the corresponding Tafel line [cf. Butler (24)]. Special Case 3. 7't less negative than - 2 0 my. From this condition, the faradaic current i", is given by (32), also from (38), in which during decay i~ = 0, and using (39) it follows that

~-7~ ],

/,

(64) (65)

io = KXF ~ - ar exp

where tAG*)1 is the standard free energy of activation for the forward direction of the rate-determining step at the reversible potential. Hence, i0 = B e x p

dT't

~io 7t F

dt Integrating (58),

CRT "

(58) where

( AH:' RT/
,

(66)
(67)

t - k/~0ff m [ 7, [ + const.

CRT ,

(59)

B = KF ~ - a, exp

Equation (59) shows that the decay curve of overpotential against time becomes asymptotic to the time axis, the relation being exponential. The equation may be used to find the differential capacity (25). (it) Relations in which dC/d7 = f(7). During decay, i" = C dT'Jdt. Using this in (38) and (39) gives
c

and AH* and AS* are respectively the heat and entropy of activation corresponding to (AG*)I. Assuming AH* and AS* are not temperature dependent, and that over small ranges of temperature B is constant, the plot of In i0 against; 1/7' has a slope of - A H * / R . Further, when n is more negative than - 7 5 my, (25) and (66) give

d7t

d,/t

(60)

ic = B e

AH*;T~TF ).

(68)

or
ir J

Hence, taking B and ~3~ as temperature independent, (61)

C = ( dd)T t ~ _ dd , h (' _ \ [t]

0(ln ic)~ _ AH* + fl~TF

(69)

-oT~] ,

R T2

where the values of C, dT't/dt, and dTt/dt refer to the same overpotential 7. Further, if (54) is applied to infinitesimal changes during decay,

From (26), and with conditions assumed as for (69),

(0,)
~o =

(70)

\d-i/,,
so that

= - - ~ exp \

RT /

(62)

From (69) and (70) it follows that (07) AH0* + ~ T F flXFT (71)

In C = Inio

RT

\ dt/,h"

(63)

2 O t h e r relations between c u r r e n t a n d t e m p e r a t u r e have been given by Agar (23).

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

176

,JOURNAL OF T H E E L E C T R O C H E M I C A L SOCIETY M~+ + ze M1 + zH,O

------+ -----)

April 1952

V. POSSIBLE REACTION PATHS OF THE HYDROGEN EVOLUTION REACTION 3

slow Ml + zOH- (L) +M1

fast M~+ + zH +OH-

A. Source of Proton is H~O+

The possible paths are (M being a metal atom): H30 + + MH + e MH e MH e

MH + e

H20 + M ~ H + e fast H2 M~+ + ze fast M~

_____~ MH + H20 fast ____, 2M + H2 fast ___~ slow ___, MH + 2M + H~O H2

slow

(A)

-----)

Ml

+ zH20

----.+

slow M~+ + zH +OH-

+ zOH- (M) +M1

H~0 + + MH +

H20 + M 1 H + e fast H2 --_~

(B)

M~+ + ze
~1

------->

fast Mx fast
)

H~0 + +
HaO + +

slow ___~. MH -~" H 2 0 fast

___, I.i~ +

"31- ZH20

M~+ -4- zH

-4- zOH- (N)

(C)
(D)

Ho.O + MtH + e slow H2

------+

+ OH- + M1

H20

H~O+ + H~0 + +

e MH +

fast - - - ~ MH + H20 slow -~- H 2 0 e ) H2

B. Source of Proton is H 2 0

H20 +

MH + MH HoO + MH + H20 + H20 + H~.O + e MI~ e MH + e e

e

____. t{2

slow , MH + fast

OH-

(E)

-~- 2M

fast _____~ MH + slow __---~ H2 +

OH2M

(F)

slow ~ MH + O H fast ____> H2 + O H - + fast ____~ M H + 0H-

(a)
M

(H)

H20 -4- MH +

slow HHo -4- OH- + M _

In alkaline solution it is also possible that the metal M ~+ ion discharge is an intermediate step in the evolution reaction. slow M~+ + ze ~ M1 M~ + zH20 MH + MH M] + + ze M~ + zH20 MH + MH M~+ + ze M1 + zH20 MH + MH f a s t M~+ + zH fast+- H~ + 2M fast
M~

The kinetics may be complicated by the presence of: (a) simultaneous reactions in which, say, two desorptive reactions proceed at the same velocity in the steady state; (b) dual reactions in which two stages ill the evolution have virtually the same energy barrier (to within, say, 3 kcal); (c) linked reactions, where the energy barriers of various steps are of appreciably different heights, but the kinetics of the overall process depend upon the heights of two or more energy barriers. It has been shown recently that discharge from water molecules in acid solutions is improbable (5). Hence, reactions E to N are likely only in alkaline solutions. The reactions A to D are the best known and the most reliable material exists by which their kinetics, developed below, may be compared with experiment. A, B, and D are commonly termed: The Slow Discharge, Atomic Hydrogen, and Electrochemical Mechanisms of the electrolytic evolution of hydrogen, respectively.
V I . THE KINETICS OF SPECIFIC REACTION PATHS

+zOH-

(I)

The most important reaction paths will be considered here in some detail. It will be assumed that the adsorption of hydrogen upon the electrode is governed by a Langmuir isotherm and that the velocity of the reverse desorption reaction may be neglected at potentials not near the reversible hydrogen potential.
A. The Source of the Protons is H~O+ and the Atomic Hydrogen Reaction is Desorptive (Paths A and B) 1. General Kinetic Equations

slow M~+ + zH f a s-t H.~ + 2M fast

M1

+zOH-

(J)

_ _ _ f s M] + + zH + zOH- (K) _ _ ~at slow ____~ H2 + 2M

The reactions and velocities Vl, v2, va to be considered a r e :

H30 + + e
___.

a The reaction paths suggested are by no means exhaustive, e.g., it is possible t h a t the discharge of hydroxonium ions may occur with an electrochemical desorption involving water, i.e., H~O + MH + e ~ H2 + O H - + M.

v~ MH,

(72) (73)

MH

- -~)2 H~O+ + e, 4

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Vol. 99, No. 4

and

CATHODIC HYDROGEN EVOLUTION REACTION

177

2. Kinetics when Discharge (Path A)

MH + MH _~
V3

is

Rate-Determining

H2 + 2M

(74)

Let al~+ be the activity of hydrogen ions in the electrical double layer in gram equivalents per liter4; a , be the activity of hydrogen atoms on the electrode surface in gram atoms per cm2; and x be the fraction of the surface covered with adsorbed hydrogen (x is more rigorously defined as the fraction of the available surface covered). Then an = l0 -9 x if there are 101~ free spaces for adsorption per cm e, of the electrode surface, and assuming the activity coefficient of the adsorbed hydrogen is unity it follows, v~ = k~ all+(1 -- X) exp

(i) General condition for slow discharge mechanism. - - T h e condition is 10(al -]- a2) < a3,

(85)

-- 2 R T / '

(75) (76)

where the factor 10 is an arbitrary limit of significance. The use of the terms "a" in the condition rather than the "k" terms provides for the possibility of changes of mechanism occurring on varying potential or hydrogen ion concentration. The linking of a~ a M a2 by a positive sign in the condition is clear, since, by the nature of the reactions concerned, a decrease in a2 increases x, thereby decreasing a~ and increasing a3: that is, a decrease in a2 increases the probability of the slow discharge mechanism, as shown by the condition (85) above. (ii) Coverage of surfaee.--From (85) and provided

v~ = /~ 10-~ x exp \ 2 ~ ] ' and va = ka 10-1Sx2,

(a~ + a2) 2 > 10a2(a~ + a~), i.e.,

al > 9a2,

(86)

(77)

where k~, k~, k3 are specific reaction velocities, and the energy barrier of the discharge reaction is assumed to be symmetrical. 3r = (Ar -- ~-) where Ar is the inner potential difference of cathode and solution and ~- is the potential difference between the bulk of the solution and the plane passing through the center of the ions adjacent to the cathode surface. In the steady state, v~ - v2 - 2v3 = 0. (78)

it is easy to show that 8a~a3 > (a~ + a2) 2. Hence using this condition in (79) it follows x = (iii) Tafel line.--From follows 9 (87)

(80), (83), and (87) it

ic = FklaH+ exp

-- 2 R T / "

(88)

Hence from (27), (29), and (30) for the slow discharge mechanism, a = 0.5; and, since X = 1, = 0.5. (90) (89)

Solving (75), (76), (77), and (78) for x gives the real solution:
X ~-

--(a, + a2) -4- g/(al + a2) ~- + 8a, a3 4a3

(79)

where al = /qan+ exp ACF '~ 2~]' (80) (81)

a2 = 10-9k~ exp \ 2 ~ / ' and a3 = 10-1Sk3. The current is, therefore, i~ = 2Fv3 = 2F10-~Sk~x ~ = 2Fa3x ~

(iv) Effect of pH in pure dilute acid solution.Assuming that the Stern model of the electrical double layer prevails at the metal-solution interface, that the solution is dilute, that specific adsorption of ions is absent, and that the interracial potential conditions are more than about 0.3 volts from those of the elcctrocapillary maximum, the following equations may be validly used, an+ = (an+)B e x p ( - ~ T F ) , ACe= A C t + n , (91) (92) (93)

(82)

(83)
and

Ar = R T In (all+).,

= F [--(al "-~ a2) -4- %/(al -4- a2)2 -4- 8ala.~]2. (84) 8a3 4 For ease of representation the suffix H30+ is replaced by H +.

t" = Const. -t- ~ '

In (aH+)B,

(94)

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

178

J O U R N A L OF T H E E L E C T R O C H E M I C A L S O C I E T Y

April 1952

where (an+)B is the activity of hydrogen ions in the

bulk of the solution. Using these equations in (88) 2RT it is found that n = Const. In i~. ThereF

Hence values of a and, therefore, of fl are complex functions of potential, and can be calculated from (101) by the use of (29) and (30), taking ~ as 2. Two important limiting conditions arise from (101).

fore, overpotential is independent of the pH of the solution with the above mechanism under the conditions (85) and (86). (v) Effect of pH in the presence of excess neutral salt.--In the presence of excess neutral salt (e.g., LaC13) ~ decreases to a constant value near zero. Under these conditions (91), (92), (93), and (94) may be combined to 77 = Const. - 2R_ lni~ -t- %T In (a,,+)B. TF (95)

Limiting condition 1 :
If k210 -9 exp > 10, (102) (103)

b - 2.303RT _ 0.029 at 20~ 2F Therefore, a = 2 a n d ~ = 1, as}, = 2

(104)

Limiting condition 2:
Equation (95) shows that under the above conditions (85) and (86), overpotential should decrease numerically by 58 mv at 20~ upon decreasing the pH of the solution by one t~nit. (vi) Effect of neutral salt at constant p H . - - F r o m (88), (92), (93), and (94), and considering (all+), as constant, ,7 = Const. If A0 tends to - ~ , b~ so that a ~ 0, and ~ ~ 0. (106) The forms of (100) corresponding to (103) and (105) are, respectively, ~, (105)

2RT In ic -- ~-.
Y

(96)

Since on addition of neutral salt to the electrolyte ~" becomes more positive and approaches zero, overpotential increases numerically.

[klaIt+12 ( ic = 2 F k s [ _ ~ - 2 J exp
and ic = 2Fkal0 -18.

2A~bF~ RT]'

(107)

(108)

3. Kinetics when Atomic Combination is Rate-Determining (Path B) (i) General condition for atomic hydrogen mechanism.--Comparing with (85), the condition is
10ca < al ~ a~. (ii) Coverage of surface.--Using condition and also the condition 9a2 < al, (97) (97) (98)

Equation (108) clearly indicates the limiting current caused by the atomic combination reaction. (iv) Effect of pH in pure dilute acid solution.Using (91), (92), and (93) in (107) it follows

RT = Const. - 2F- In ic.

(109)

and also since %/1 ~- n = 1 + 89 where n is small, it follows a, x . (99) al + 62 (iii) Tafel line.--From (80), (81), (82), (83), and (99), i ~ - - 2F10-18ks al
a l -~- 62

Equation (109) indicates that with the above mechanism overpotential is independent of the p H of the solution. It is evident from (108) that the magnitude of the limiting current is independent of pH. (v) Effect of pH in the presence of excess neutral salt.--Equation (109) shows that ~ is independent of ~" and, therefore, is independent of pH in the presence of excess neutral salt. Equation (108) shows that the limiting current is independent of the presence of neutral salt. (vi) Effect of neutral salt at constant pH. ~ is unaffected by the presence of neutral salt (see 109).

---- 2Fksl0 -1~

k~lO- i A4~F 1 -I- ~a~H+ exp ~

4. Minimum Coverage of Cathode by Hydrogen for Atomic Combination to be Desorptive

From (16), (77), and (83), ic 2F ~T 10-18x2, (110)

d(A~) Evaluation of d(ln ic) yields d(A~b) _ d(ln i~) b 2.303 _ 1 + klk2 aH--------~ exp \~][A~bF~ 10-9 2Fk~ 10-9 (A4~F~ " RTklaa exp \ R T ]

(101)

if (AG*)I is taken as zero to give a maximum velocity of combination, ~ = 1, and kl in (16) is identified

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Vol. 99, No. 4

CATHODIC HYDROGEN EVOLUTION REACTION

179

with ka in (83). Hence, the least possible coverage for the atomic hydrogen reaction to maintain a current density of i~' a m p / c m ~ is given by x = 10~ . (111)

(it) Coverage of surface.--Utilizing (118) in (114), and also from (80) and (115),

x = a~/a4 = k~/lO-gk4.

(119)

(iii) Tafel line.--Condition (118)utilized in (116) gives, ic = 2FklaH+ exp ACF~ 2RT]' (120)

B. The Source of Protons is H~O + and the Electrochemical Reaction is Desorptive (Paths C and D) 1. General Kinetic Equations
The reactions and velocities to be considered are HaO + q- e MH and
HaO + +

- - - ~ MH, __v~_. HaO+ + e,

Vl

(72) (73)

which relation is very similar to that found for a rate-determining discharge reaction followed by an atomic hydrogen desorption step. Hence, a = 0.5 andfl = } a s h = 2. (iv) Effect of pH in pure dilute acid solution, (v) effect of pH in presence of excess neutral salt, and (vi) effect of neutral salt at constant pH.--These three effects are as already given in Section A, 2.

M H -b e ----* H~.

V4

(112)

3. Kinetics when Electrochemical Step is Rate-Determining (Path D)

(i) General condition for electrochemical mechanism.--The condition is 10a4 < al -b a2. (it) Coverage of surface.--Special case (a) 10al < a2. From (114), (121), and
X =

v, and v2 are given by (75) and (76). v~ is given by

v~ = tc~lo-ga~+x exp

-- 2RT]"

(113)

(121) (122)

The steady state value of x yielded by reasoning analogous to that leading to (79) is x where
a4 =

(122),
at~a2.

al al -~- a2 -~- a4'

(114) Special case (b)

(123) (124)
(125)

10a2 < k410-ga~* exp

at.

-- 2RT]"

(115)

From (114), (121), and (124),

x -~ 1.

The current is, therefore,

(iii) Tafel line.--Special case (121) and (122). From (116), z~4~F'~2RT]
9 (116)

(a).

Conditions

2Fk~k4 10-9(all+) 2 exp (

ic = 2F k~ k~(a~+)2 exp

(k~ -b k410-~)a.+ q- k210 9 exp \ ~ ] From (116),

d(a~) d(ln ic)

(126)

From (27), (29), and (30) it follows that a = 3/2 as }, = 2,

k21o-~ (~F~ (/c~ ~ k410-9)a~+ exp \ R T ]

(127) (128)

fl -- 3/4.

1+

(iva) Effect of pH in pure dilute acid solution.Using (91), (92), (93), and (94) in (126) it follows (117)

2RT
-

1 + (k~ ~- k4 10-9)an+ exp \ ~ - ] _ ]

2RT = Const. - - - In G. 3F

(129)

(k~ + k, IO-9)aH+RT exp

\-~/3

2. Kinetics when Discharge is Rate-Determining (Path C)

(i) General condition for slow discharge mechanism. - - T h e condition is 10(a~ + a2) < at. (118)

Hence with the above mechanism under the above conditions overpotential is independent of the pH of the solution. (va) Effect of pH in the presence of excess neutral salt.--From (91), (92), (93), and (126) and as ~" -~ 0 it follows that: = Const. - 2RT In ic q- ~ -T In (aH+)s. R 3-F(130)

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

180

J O U R N A L OF T H E E L E C T R O C H E M I C A L SOCIETY

April 1952

Equation (130) shows that under the above condition overpotential decreases numerically by 19 mv at 20~ upon decreasing the pH of the solution by one unit. (via) Effect of neutral salt at constant p H . - - F r o m (91), (92), (93), and (126) and considering (a.+)B as constant, 2RT -- Const. - 3 F - In i, - ~'/3. (131)

In both reaction schemes desorption of hydrogen may occur by means of the reactions, HgH + HgH ~ H2 + 2Hg H20 + HgH + e ~ H: + OH-. (f) (g)

Hence, as ~" becomes more positive and approaches zero on addition of neutral salt, overpotential increases numerically. (iiib) Tafel line.--Special case (b). Conditions (121) and (124). From (116), io = 2FkaaH+10 -9 exp 2RT]"

It is improbable that reaction (a) occurs because evidence is available (7) that alkali metal deposition occurs on Hg without being accompanied by an appreciable overpotential. Reaction scheme cdef will, therefore, be formulated. It can be shown that very similar results are obtained for the scheme cdeg (6). Let v5 be the velocity of reaction c, v~ that of d, v7 that of e, and Vs that of f. Let Xl be the fraction of the surface of mercury covered with alkali metal atoms in the steady state and x2 be the similar fraction for hydrogen atoms. Then, v5 = ksan+(1 -- X l - x2) exp

From (27), (29), and (30) it follows that a = 0.5 as h = 2, /3 = 1/4. (133) (134) v6 = k610 -9 xl exp \ 2 R T - ] '
vr = k7 ( 1 0 - 9 Xl) t a H 2 0 ,

2~]'

(135) (136) (137)

(ivb) Effect of pH in pure dilute acid solution, (vb) effect of pH in the presence of excess neutral salt, (vib) effect of neutral salt at constant p H . - Consideration of (91), (92), (93), and (132) shows that the above three effects are as already given in Section A, 2. C. The Source of the Protons is Water and a Metal Cation Takes Part in the Reaction: A Special Case Reference to the reaction paths E to N indicates that many more possibilities exist for the mechanism of the evolution of hydrogen when the source of the protons is water than when it is the hydroxonium ion. Since little data as yet exist with which comparison m a y be made, the kinetics of these schemes will be formulated here only in one special case. It has been recently shown by Bockris and Watson (6) that the evolution of hydrogen at mercury cathodes from aqueous alkaline solutions involves the alkali or alkaline earth metal cation. Hence, the possibilities in this case are: M ~+ + ze slow ~ M/Hg (a) + z H g H + z 0 H - (b)

where xi is raised to the power 89in accordance with the experimental data of Jofa and Pechkovskaya (8) on the velocity of reaction between potassium amalgam and water, aH~o is the activity of water, and ks-kr are velocity constants. At low and medium current densities, l >> xl, x2. Also, in the steady state, v5 - v6 - - vr = 0,

(138) (139)

v7 -- Vs = 0. According to the mechanism suggested, k~all+ exp 2RT ] >> k710- 9/2 aH2o

(140) <~ kG10-9 exp \ 2 R T - ] " From (140) by an argument similar to that used in the derivation of (88), one obtains (k4 89 ~ (z~F~ io = Fk7 \k6] aH20(aM*+)~ exp 2RT-]" But,
k5

(141)

ke

exp

_ (aG*)~- (aG*)~)
R-T = exp

M / H g + zH20 f_~ast M,+ or, M ~+ M/Hg M/Hg+ + ze fast ---+M/Hg f a s t M~+ zH20 sl~

(142) -- R T ] '

(c) +ze +zHgH+zOH(d) (e)

where AG5 is the standard free energy change in reaction (c). Because the departure from irreversibility in (135) and (136) is small it follows that approximately,
AG5 =

-- ze~oF ,

(143)

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Vol. 99, No. 4

CATHODIC HYDROGEN EVOLUTION REACTION

181

where e'0 is the reversible electrode potential for the alkali metal concerned at unit activity in mercury. Hence, from (141) and (142) and (143), ir = zFau~o (aM~+)' k7 exp 2~ [e0 -- A4] . (144)

B. Numerical Calculations 1. General Remarks and Preliminary Calculations

The results of calculations presented here are to be regarded as of order accuracy only. This is due to (i) the approximation involved both in the equations themselves and in the method of numerical application. (e.g., calculation of reversible eleetrode potential, see below), and (ii) the inaccuracy of some of the experimental and theoretical data. The following results are intended to be exemplify-

A reasonably good value of e~ can be obtained from the emf of a reversible amalgam electrode of the type M amalgam/M ~+ ions at unit activity. If it is assumed that the interaction energy between M and Hg is included in e~, and that it is constant

TABLE I. Characteristics of various mechanisms of cathodic hydrogen evolution

Wi $yl metr cat[ enel gY I X ban mr I LO(~nn

Characteristic of mechanism

Conditions for application

aH+)BAr, i

Neutral salt effect at constant pH + Excess neutral salt 1

--v

ol
--i Acid solution mech- Slow discharge anisms F a s t atmnie H (A) Slow atomic H F a s t discharge (B) Slow discharge F a s t electrochemical (C) --i Slow electrochemical a, + a 2 > 10a4[a~ > 10m 10(aL + a2)< at 9a2 < at 10as < al + a~ 9a~ < al 10(a~ + a ~ ) < a~ 2 2.1 I
m

Pure solo. Nil

RT

~ more neg.

1 ;

Nil

Nil

Nil

Nil

1 --~

RT F

~ more neg.

2 2

I I1
I

Nil

1 -T l+v 1 -'y

RT F RT F

~ more neg.

F a s t discharge (D)

'al > 10a2

t I1
--i

Nil

~ more ncg.

Alkaline solution meehaIlisms

Slow discharge from Htl3 w i t h either fast atomic n (Eb or fast electrochemical from H~O* (G) Slow M / H g ) reaction F a s t M z+ discharge F a s t atomic H ( J )

Analogous to m e c h a n i s m s with discharge from | h O +

--2RT
Y

--RT Y

~ more pos.

as > 10aT < a e

--RT RT ~ - 47 ~ -

0 [n(a~u z+) B 9 0 ln(a,+) B If ~ = 0.5

--RT -Y-

Nil

~, = Bttrrier d i s y m m e t r y factor = Fraction of potential assisting forward direction of rate-determining step. * Assuming u n i v a l e n t alkali metal ion. ~: A s s u m i n g t h a t relation ~- = const 4- ~ In (an+)B is valid.

over a small change of activity, then the electrode potential e, is given by , ea = e0 + - 1 In ~ ,

(145)

where C M ---- concentration of alkali metal M in Hg. From (145), knowing e, experimentally, e'0 is found.
VII. DIscussioN OF MECHANISM AT Hg, Ag, Ni, AND Pt CATHODES

A. Table of Distinguishing Criteria

These are given in Table I which is based upon the general results to be found in Sections IV and VI.

ing rather than exhaustive (e.g., throughout, the concentration of hydroxonium ions in the electrical double layer has been taken to be 10-1 gram equivalents per liter and no quantitative examination has been made of the effect of variation of concentration). Evaluation of terms such as exp (AeorF)/RT has been made by expressing A4: as the potential of the reversible hydrogen electrode with respect to the eleetroeapillary maximum of the metal concerned. The approximations involved in this procedure, involving the neglect of the potential difference at the interface due to adsorbed solvent dipoles, should be recognized; in comparison of rates on various metals, this error is less serious because it is

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

182

JOURNAL OF THE ELECTROCHEMICAL SOCIETY

April 1952

reasonable to assume that the dipole contributions are approximately the same at different metals. The terms AG~*, AG2*, AG*, and AG* (where AG* is a standard free energy of activation; 1 and 2 refer respectively to the forward and reverse directions of the discharge of hydroxonium ions; ~nd 3 and 4 refer respectively to the forward atomic hydrogen and electrochemical desorption steps) have been obtained as follows. AG* and AG* have been taken from calculations by Parsons (26) of the energetics of the discharge reaction 5. AG* (for the atomic
T A B L E II. Certain numerical
Reversible I H poten- I Log i0 ]tiai referred AG~* (aa+ = to electro- kcal 0.1) capillary maximum volts

Table II gives numerical values of some of the quantities used in the following calculations.

2. Potential Condilions for Applicability of Slow Discharge Mechanism (Path A)

The two conditions to be satisfied for (88) to apply are (85) and (86). Numerical application of these conditions gives the results of Table III.

3. Coverage of Surface (x) for Slow Discharge Mechanism (Path A)

Table IV gives the surface coverages at various overpotentials obtained by applying equation (87) at 20~

uantities at 20~

Cathode

Log k~

AG~* kcal

Log k~

k~/k2

4. The Value of AG* for Atomic Hydrogen Desorplion

For smooth Pt cathodes the observed experimental slope of the Tafel line at current densities of about 10 2 _ l0 ~ amp/cm 2 indicates that the atomic hydrogen mechanism is rate-determining. The limiting current density for this mechanism is found to be 30 amp/era 2 in 1N aqueous acid solution (13). Substituting in (146) it follows that, AG* = --2.5 kcal approx. k3 = 1014'2 The correct value of AG* can therefore be taken as about 0 kcal, the corresponding value of k3 being 10'2's at ordinary temperature. With the approach of saturation appreciable departures from ideality among the hydrogen atoms on the surface are likely to occur. Hence, the react'ion may be somewhat slower than indicated by the value of AG*. A low value of AG* would be expected because the reaction concerned is a surface radical reaction. To a first approximation it is assumed that AG* (and also k3) are the same for all cathodes.

Hg Ag Ni

-12 -7 -6

+0.20 38 --0.046 31 --0.30t 31

-15.5 -10.3 -10.3

23.2 23.9 23.6

-4.5 --5.0 -4.7

--11.0 - 5.3 - 5.6

Value o b t a i n e d from i n t e r p o l a t i o n of an observed relation between 9 a n d the p o t e n t i a l of the electrocapillary m a x i m u m for a n u m b e r of metals. T A B L E I I I . Potential conditions for slow discharge mechanism (Path A) at 20~
Cathode Condition (85) : Overpotential in volts Condition (86) : Overpotential in volts

Hg Ag Ni

More positive t h a n - 1.4 More positive t h a n - 0 . 5 4 More positive t h a n - 0 . 2 9

More negative t h a n -0.43 All v a l u e s ~ s a t i s f y condition All values satisfy condition

T A B L E IV. Surface coverage with hydrogen at various overpotentials (Path A)

Cathode Hg Ag Ni Overpotentia] (volts) Coverage x Coverage x Coverage x --0A l0 -6"2 10-2"6 10-1'5 --0.3 10-a'4 10-L~ 10-~ -0.5 10-4.5 10-~ --0.7 10-3.4 --1.0 10-2"4 --1s

5. The Value of lcl/k2 for Pt Cathodes

Sat, d. Satd. surface. Saturated surface

Utilizing the ahove determined value of AG* (or k3) in (100), (80), and (81) referred to the apecial case of the exchange current i0 so that '1 = 0, it follows that
kl/k2 =

hydrogen desorption) will be shown below to have a value of approximately zero. AG* (for the electrochemical desorption) cannot yet be calculated even with sufficient accuracy for the present purposes; k~, where n = 1, 2, or 3 depending on the reaction concerned, has been calculated from AG* by means of the equation, k, = ~ - exp -

10 -14"4.

6. Calculation of the Course of the Tafel Line for the Pt Cathode

Substituting the above obtained values 8 for k3 and kl/k2 in (100) gives the complete Tafel equa6 In this calculation the experimental value of ks i.e. 10 L4-2,has been used in order t h a t the l i m i t i n g c u r r e n t shall appear at 30 a m p / c m 2. If the value of k3 = k T / h = 10~* is used the l i m i t i n g c u r r e n t appears a t 1 a m p / e r a 2. In e i t h e r case the form of t h e calculated Tafel line remains the same. T h e values of log i0 and the reversible h y d r o g e n p o t e n t i a l referred to the electrocapillary m a x i m u m h a v e been t a k e n as - 3 , a n d +0.28 volts, respectively.

RT]"

(146)

All numerical values q u o t e d i n following sections hence depend upon the accuracy of these very a p p r o x i m a t e calculations, except for values calculated here concerning platinum.

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

VoL 99, No. 4

CATHODIC HYDROGEN

EVOLUTION REACTION

183

tion for hydrogen evolution from aqueous 0.1N HC1. The two limiting slopes deduced in (103) and (105) are shown (see Fig. 3).

9. Conditions for Applicability of Slow Discharge Mechanism (Path C)

The condition to satisfy for (120) to apply is (118). However, in this case it is not possible to calculate sufficiently accurate values of k, or AG* for potential conditions to be found. Limiting values for k, and AG* when P a t h C is rate-determining have therefore been calculated and are given in Table V I I . Calculations of AG* b y statistical mechanical means are complex and have only been a t t e m p t e d by Conway (9) in the case of Ag. He finds (hG*)^~ 9 kcal, in which case, therefore, discharge would be rate-determining if the desorption were electrochemical. TABLE VII. Limiting values of k4 and AG*~for Path C to be

7. Minimum Coverage of Cathode by Hydrogen for the Atomic Hydrogen Mechanism to be Desorptive
The required m i n i m u m coverage is obtained from ( l l 1) as a function of current and is given in Table V.
-020
] J l r

-- 01 1 ~

-0,1(

rate-determinin at 20~
Cathode
+005

k4 limit for Path C, Rate-determining

AG* limit for Path C, Rate-determining

~ "~
-3.0 -2 0

,
- IO

,
0

,
1.0 2.0

Hg

When ,1 more negative When ,1 more negative than --0.43 volt aG~ than -0.43 volt k, must be < 24 kcal must be > 10 -5.~ k4 must be > 1O -~ at ~G~ must be < 17 kcal at all overpotentials all overpotentials As for Ag As for Ag

log Ic

Ag Ni

FIG. 3. Calculated course of Tafel line for the smooth platinum cathode in 0.1N HC1 at 20~ (assuming path B is rate determining). TABLE V. Minimum coveragefor atomic hydrogen desorption
ic amp/cm~ 10-t~ x

TABLE VIII. Maximum values of surface coveragefor Path

C to be rate-determining
Cathode Maximum value of x

lO-g

10-6 10-3
1

10

10-8 10-4.5 10-3 lO-i .6 1 (satd. surface) Saturated surface

Hg Ag Ni

10-~ at overpotentials more negative than -0.43 volt 10-' at all overpotentials As for Ag

TABLE VI. Potential conditions for atomic hydrogen mech-

anism (Path B) at 20~

Cathode Condition (97) .Overpotential in volts . . [ . Condition (98) Overpotential in volts .

10. Coverage of Surface for Slow Discharge Mecha-

nism (Path C)
The coverage x is given b y (119), and by using the limiting values of k4 given in Table V I I a maxim u m value of x can be calculated (see Table V I I I ) . 11. Conditions for Applicability of Electrochemical

Hg Ag Ni

More negative than' More positive than -1.65 [ -0.43 More negative than I All values satisfy con-0.78 than dition More negative All values satisfy con-0.64 dition

Mechanism (Path D)
The condition to satisfy for P a t h D to be ratedetermining is (121), b u t since k4 and AG* are not known accurately, Table I X gives limiting values of these quantities (cf. Subsection 9). 12. Potential Conditions for Applicability of a =

8. Potential Conditions for Applicability of Atomic Hydrogen Mechanism (Path B)

The two conditions to be satisfied for (100) to apply are (97) and (98). Numerical application of these conditions gives the results of Table VI. For smooth platinum cathodes, the use of the experimental values of k,/k2 in condition (97) shows t h a t for the atomic hydrogen mechanism to apply a t ~ = 0, kl m u s t be > 1 0 -6"s.

and a = 89when Path D is Rate-Determining

The condition to be satisfied for a to have values of ~ and 89 are, respectively, (122) and (124). The potential conditions appropriate to both values of are given in Table X.

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

184

JOURNAL OF T H E E L E C T R O C H E M I C A L SOCIETY

April 1952

13. Coverage o.f Surface for Electrochemical Mechanism (Path D) The required coverage is found from (114) neglecting a4 and is as follows: Hg: Surface saturated at overpotentials more negative than - 0 . 3 7 volt, Ag:~ Surface saturated at all overpotentials. Ni: )

C. Estimate of Rate-Determining Reactions at Hg, Ag, Ni, and Pt Cathodes According to Evidence at present Available 1. Mercury
(i) In Acid Solution.--The value of a -- 89shows that the mechanism must be A, C, or D. At low current densities the velocity of the atomic comT A B L E I X . Limiting values of k4 and 5G*4for Path D to be

rate-deterndning at 20~
Cathode k4 limit for Path D, Rate-determining AG~'limit for Path D, Rate-determining

Hg

When n more negative When n more negative t h a n - 0 . 4 3 v o l t k4 t h a n - 0 . 4 3 v o l t AG~ m u s t be < 10-7-~ m u s t be > 27 kcal k4 m u s t be < 10 - 2 3 a t all o v e r p o t e n t i a l s AG~ m u s t be > 20 kcal a t all o v e r p o t e n t i a l s

Measurements in this current density region have been made (12) and do indeed show an increase in the value of a. Here, however, the continuous increase of a with decreasing current density probably indicates the presence of slight trace impurities in the mercury used (to which traces the kinetics of the electrode reaction would be very sensitive at the very low current densities involved). The mechanism is hence probably A or C. According to the results of Sections (VII, B, 4 and 8), the mechanism cannot be A above a current density of about 30 amp/cm 2 and below about 10-s amp/era 2, outside which limits it must therefore be C. It is not possible to distinguish at present between A and C at intermediate current densities although for both the rate-determining reaction is the discharge step. (ii) In alkaline sohdion.--As shown in the special case worked out in Section (VI, 0), the rate-determining reaction here is tile combination of alkali metal atoms with water. An implication of this mechanism is that it would give rise to a limiting current density at high current densities, but this aspect is as yet unexamined.

2. Sih, cr in Acid Solution

The value of a (9, 13) shows that the mechanism must be A, C, or D. The results of Section (VII, B, 2) show that mechanism A cannot occur at overpotentials more negative than - 0 . 5 4 volt, ~nd furthermore is unlikely at high current densities since no limiting current is detectable up to current densities of 150 amp/cm 2 (13). If the mechanism is D, Section (VII, B, 12) shows that a must equal 1 as is observed. A determination of h (14) yields the value 1 which favors mechanism A ; and statistical mechanical calculations by Conway (9) have shown that 5G~ is about 9 kcal, whereas for mechanism D to occur AG~ must be >20 kcal. Effect of pH is complex but can be brought into better accord with A or C if Stern's equations for the dependence of electrokinetic potential on concentration are developed fully.

Ag

Ni

As for Ag

As for Ag

TABLE X
Cathode Hg __C~176 for a = 3/2_ n more positive than --0.32 v o l t N o v a l u e of ~ N o v a l u e of n Condition for a = 1/2 I ~ more negative [ -0.43 volt ] All v a l u e s of I All v a l u e s of I than

Ag
Ni

bination is too great for mechanism D to occur (see Section VII, B, 11). I t is therefore highly improbable that this mechanism occurs at all on mercury because a change to it at higher current densities would involve a change in the value of i0 and this is not observed. Further, according to the conclusion of Section (VII, B, 13), if D were operative, the surface of the mercury would be saturated with adsorbed hydrogen at all potentials accessible to experiment, and this is contrary to evidence concerning the variation of the interracial tension of mercury-solution interfaces with pH (11). According to the results of Section (VII, B, 12), mechanism D with a value of a = ~ would occur up to an overpotential of - 0 . 3 volt (corresponding to a current density of about 10 9 amp/era 2) if electrochemical desorption were the rate-determining reaction.

3. Nickel
(i) In acid solution.--The value of a showsthat the mechanism must be A, C, or D (9, 10, 15). Determination of ~, is difficult since the Tafel line is of the dissolution type (see Fig. 1) and at intermediate current densities in strongly acid solutions ( > N / 2 ) the dissolution of the metal partially obscures pH and salt effects. Azzam and Bockris have shown (13) that a limiting current occurs at about 1 amp/cm 2 which supports mechanism A. The absence of pH effects at concentrations below N/100 also supports this mechanism. The results of Section

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Vol. 99, No. ~

CATHODIC HYDROGEN EVOLUTION REACTION

185

(VII, B, 2) show that mechanism A cannot take place at overpotentials more negative than - 0 . 3 volt, and it is interesting to note that the Tafel line begins to attain a limiting current at about this potential when measurements are made using a fresh cathode surface (10). (iN) In alkaline solution.--The coefficient a is 89and )` is observed to be 1 (10), which indicates that mechanism E or G is operative. This is confirmed by the interpretation of the observed pH effects. 4. Smooth Platinum in Acid Solution The value of a at intermediate current densities indicates that the mechanism is B (16). The mechanism at current densities above the limiting current density may be C or D, but is probably D owing to the saturation of the surface with adsorbed hydrogen at the limiting current density.
VIII. SUMMARY OF SOME RECENT ADVANCES

The controversy concerning the value of b in the Tafel equation now seems resolved experimentally as follows. If the solutions are pure, b is not (except transiently at a limiting current) greater than about 0.13, i.e.,/~ > 0.45/),. This fact confirms that the energy barrier for the discharge reaction is usually nearly symmetrical and is in accordance with the equations formulated in Section (VI). Higher values of b are probably connected with a distortion of the energy barrier (the position of the transition state complex being displaced toward the electrode) due to the specific adsorption of poisons on the electrode. The realization of the use of )` as a distinguishing criterion of mechanism, and its intensive statistical application, has made available a criterion of greater power than previously existed. Knowledge of AH0* and the potential of the electrocapiUary maximum aid application of the quantitative conditions established above. In recent years, it has been suggested (17) that prototropic transfer of hydrogen from water in the solution to water adsorbed on the electrode, and also (18) that the atomic hydrogen desorption replaced at higher current densities by electrochemical desorption are the main rate-determining reactions to be considered for the hydrogen evolution reaction. Evidence in favor of the first was based mainly upon the constancy of the factor B for various electrolyte concentrations and for a few metals. Consideration of (67) shows t h a t the first requirement would be expected for any mechanism, however, and it has in any case been shown in an earlier section of this paper that a sufficiently accurate ex-

perimental determination of B is prohibitively difficult [see also Butler (19)]. In order that the prototropic transfer theory can yield a satisfactory theory of pH effects and of reversible electrode potentials an unusual structure of the double layer has to be assumed (20). Forecasts made upon this basis of similarities between hydrogen and oxygen overpotentials have also not been confirmed (27). The second theory (18) suffers the fundamental disadvantage that it involves the formulation of the kinetic processes at working hydrogen electrodes in terms of equations applicable only under equilibrium conditions; it is impotent in indicating expected pH and salt effects; and the explanation suggested for different values of b is based on experimental results which are not in accord with those obtained using very pure solutions. Lastly, a quantitative, statistical mechanical formulation of the discharge reaction from both hydroxonium ions and water has recently been made (5). This work indicates rates of the discharge reaction in good accord with those experimentally observed on the two electrode materials considered (Hg and Ni). There also appears to be only a very small probability that discharge can proceed from water molecules in acid solution. ACKNOWLEDGMENT Thanks are due to Dr. R. Parsons for critical discussion and to Dr. R. G. H. Watson for verification of the numerical calculations.
Any discussion of this paper will appear in a Discussion Section, to be published in the December 1952 issue of the
JOURNAL.

REFERENCES
1. J. O'M. BOCKRIS AND B. E. CONWAY, Trans. Faraday Soc., 45,989 (1949). 2. A. M. AZZAM, J. O'M. BOCKRIS, B. E. CONWAY, AND H. ROSENBERG, Trans. Faraday Soc., 46, 918 (1950).
3. J. O'M. BOCKRIS AND n . E. CONWAY, J. Sci. InsIru-

ments, 25,283 (1948). 4. J. O'M. BOCKmS, Chem. Revs., 3, 561 (1948).

5. R. PARSONS AND 3. O'M. BOCKmS, Trans. Faraday Soc.,

47,914 (1951). 6. J. O'M. BOCKmS AND R. G. H. WATSON,J . chim. phys., 49, 1 (1952). 7. J. HEYROVSKY, Disc. Faraday Soc., l , 212 (1947). 8. J. JOFA AND Z. B. PECHKOVSKAYA, Doklady Akad. Nauk. S.S.S.R., 59,265 (1948). 9. B. E. CONWAY,Thesis, University of London (1949). 10. J. O'M. BOCKRIS AND E. C. POTTER, J. Chem Phys., March (1952). 11. A. N. FRVMKIN, Acta Physicochim. U.R.S.S., 18, 23 (1943). 12. F. P. BOWDEN ANn K. E. W. GaEW, Disc. Faraday ~oc., 1, 86 (1947). 13. A. M. AZZAM AND J. O'M. BOCKRIS, Nature, 165, 403

(t95o).

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

186

JOURNAL OF THE ELECTROCHEMICAL SOCIETY

April 1952

14. E. C. POTTER, Unpublished results. 15. P. LUKOVTSEV,S. LEWINA, AND A. N. FRUMKIN,Acta Physicochim. U.R.S.S., 11, 21 (1939). 16. A. M. AZZAM, Thesis, University of London (1949). 17. H. EYRING, S. GLASSTONE, AND K. J. LAIDLER, J . Chem. Phys., 7, 1053 (1939). 18. A. HICKLING AND F. W. SALT, Trans. Faraday Soc., 38, 474 (1942). 19. J. A. V. BUTLER,J. Chem. Phys., 9,279 (1941). 20. G. E. KIMBALL,S. GLASSTONE, AND A. GLASSNER,J . Chem. Phys., 9, 91 (1941).

21. J. HORIUTI AND M. IKUSIMA,Proc. Imp. Acad. Tokyo, 15, 39 (1939). 22. B. KABANOV AND J. JOFA, Acta Physicochim. U.R.S.S., 10, 616 (1939). 23. J. N. A6AR, Disc. Faraday Soc., 1, 81 (1947). 24. J. A. V. BUTLER, Trans. Faraday Soc., 28, 379 (1932). 25. P. LUKOVTSEVAND S. LEWINA, J. Phys. Chem. U.S.S.R., 21,599 (1947). 26. R. PARSONS, Z. Elektrochem., 55, l l l (1951). 27. A. HICKLING AND S. HILL, Disc. Faraday Soc., 1, 236 (1947).

Downloaded 30 Aug 2010 to 165.91.74.118. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp