A Phenomenological Theory of Electrosorption

Gary Anthony Attard

Department of Physics

The Oliver Lodge Laboratory

University of Liverpool

Liverpool L69 7ZE

e-mail: attard@liv.ac.uk

ABSTRACT

A phenomenological theory of electrosorption is presented based upon notions of local

charge and local electronic polarisability (Friedel oscillations). The strong link between the

extent and nature of ionic and molecular adsorption at an electrode surface and the uv-

absorption properties displayed by such molecular entities in an aqueous solution is

emphasised. The theory may be used to predict, in certain cases the extent of ionic

adsorption at the potential of zero charge (PZC) and, also, to rationalise recent results

pertaining to metal underpotential deposition (UPD) processes on single crystal electrodes.

1.

Introduction

The precise nature of the interphase between an electronic conductor and an

electrolyte remains, even after almost a century of intensive scientific research, the central

problem in electrochemical science [1]. This assertion is made in the full knowledge that the

distribution of charges and solvent molecules within the interphasial region and their

dependence upon potential, charge, pH and ionic strength will fundamentally determine all

measurements/transformations carried out via an electrochemical process. However, even a

cursory glance at the copious amounts of literature relating to the thermodynamics and

kinetics [2-5] of electrified interfaces will reveal that the situation is not straightforward.

One of the most contentious issues arising from these studies may be summarised as

follows:-

What are the forces involved in the specific adsorption of ions and, in particular,

how are these ions bonded to an electrode surface?

A working definition of specific adsorption has been put succinctly by Bockris and

Habib:-

"The specific adsorption at the interface between an electronic and ionic conductor is

the adsorption which is in excess or deficit of the amount which would be expected to be

present in the interface from simple Coulombic considerations [1]."

That is, there is an added dimension to the bonding involved in specific adsorption

over and above any electrostatic interaction and which, as yet, remains unresolved. In this

paper, an attempt will be made to address some of the difficulties which exist concerning

physical models of specific adsorption and its major themes, of ionic hydration [6 - 7],

"contact" adsorption [8], partial charge transfer [9 - 10] and donor-acceptor properties of

metal-ion pairs [11].

This paper will be in three parts: firstly, an exposition will be given of a new model

of adsorption based upon local electronic polarisability and local charge; predictions based

2

upon this model will then be tested against past, present and newly-emerging data pertaining

to specific adsorption. Second, the consequences of extending the model in a general way to

adsorption processes associated with underpotential deposition will be examined. Finally, the

relative success of the new model for the interpretation and prediction of electrosorption

behaviour with respect to presently-accepted notions of specific adsorption will be discussed.

It is hoped that a coherent, semi-quantitative picture of adsorption at electrode surfaces may

thus emerge.

2. Experimental

A limited number of electrochemical experiments were performed in support of the present

study. These involved cyclic voltammetric measurements of Pt{111} and PtPd{111} bulk

alloy single crystal electrodes. The experimental apparatus together with the preparation

procedures used for manufacturing single crystal electrodes have all been described

previously as has the sources and purity of all reagents used [12 - 14]. All electrolytes were

prepared using pure water obtained from a Milli-Q water system [12] and the flame-

annealing method of Clavilier [15] was utilised to produce clean, well-defined electrode

surfaces. All copper underpotential deposition measurements used a copper wire in contact

with the electrolyte as a reference electrode.

3. Results and discussion

3.1 The Basic Elements

3.1.1 The ion

Figure 1 (a) shows schematically what happens when a negative point charge

approaches a metallic conductor [16].

3

(a)

(b)

4

(c)

Figure 1: (a) Charge oscillations as a result of a test charge approaching a metal surface

in vacuum. (b) The same situation assuming electronic substrate mediated interactions

for ions approaching an electrode surface. (c) Enhancement of amplitude of Friedel

oscillation for case of cations and anions. The ovoid shape of the iodide and caesium

ions represents the magnitude and direction of the polarisation induced by the local

charge at the surface. The magnitude of the local Friedel oscillation amplitude is in turn

represented by the sign and size of the ‘segment’ in (b) at the surface.

As the test charge approaches the metal surface in vacuum, charge oscillations are

established extending both across the surface and into the bulk and these are referred to

as Friedel oscillations [17 - 18]. In an analogous situation whereby an anion approaches

a metal electrode, because of the dipoles and monopoles that are present in the electrical

double layer, it may be envisaged that the amplitude of the Friedel oscillation at the

surface may be enhanced/stabilised by these species as depicted in figure 1(b) in which

charge displacement takes place laterally across the surface of the electrode with

polarised water dipoles or monopoles stabilising displaced charge. The concept of

electronic substrate mediated interactions is already well established in surface science

studies of adsorption at single crystal surfaces [19]. Similarly for a cation, the same

configuration may be attained but with the polarised water molecules orientated in the

5

opposite direction (Figure 1 (c)). This may be thought of as a monopole or dipole

stabilised Friedel oscillation.

In this representation, the degree of ion adsorption will depend upon three factors:

(i) the polarisability of the ion;

(ii) the ability of the polarised solvent molecule to stabilise displaced charge;

(iii) the ability of the metal to localise charge at the ion adsorption site (amplitude of the

Friedel oscillation).

Factor (i) is a function of several variables but both ab initio calculations [20 - 23] and

experimental measurements [24 - 27] of the polarisability of halide ions

have demonstrated that, indeed, the polarisability of the halides increases in the order:

F- < Cl- < Br- < I-

which also happens to reflect the strength of specific adsorption displayed by these ions.

However, when oxyanions are considered, the correlation of specific adsorption with

polarisability breaks down. According to experiments using well-ordered Pt{111}

electrodes, the relative shift in the potential of the "butterfly peak" [28] is a function of the

strength of specific adsorption. If this is correct, the order of specific adsorption should be:

ClO4−

≈ F- < HSO4−

≤ H2PO4−

< C1- < Br- < I- …….(1)

Similarly, calculations of the polarisabilities of the oxyanions in the gas phase indicate that

they exhibit far greater polarisabilities than gas phase chloride ions [29 - 30], as expected

since oxyanions contain many more constituent atoms! Hence, specific adsorption seems not

to be explicable in terms of simple molecular polarisation. However, if one assumes that the

total polarizability of an oxyanion is comprised of individual contributions from each of the

N-O, S-O and P-O bonds (i.e. a localisation of polarizability defined by total polarizability

divided by number of N-O, S-O or P-O bonds) then the values obtained do fit the

electrochemical trend [29]. One of the factors which does influence the polarisability of a

molecule or ion is the difference in energy between the highest occupied molecular orbital

6

(HOMO) and the lowest unoccupied molecular orbital (LUMO) [30]. A proxy for this

energy difference is the uv-absorption maxima of aqueous ions.. Therefore, it was thought

interesting to plot the specific adsorption of the halides [31] versus their uv absorption

maxima in aqueous solution [27] (Figure 2).

Figure 2: A plot of specific adsorption of anions at mercury electrodes as a function of uv-

absorption maximum of the ion. Data for y-axis obtained from references 24 – 27, 32 - 33

and the x-axis data taken from reference 31. All uv-adsorption maxima relate to charge

transfer to solvent transitions (CTTS).

As expected, a monotonic increase in specific adsorption is observed as νmax increases. If one

now uses this curve to read off the specific adsorption of bisulphate and dihydrogen

phosphate ions based on their uv-absorption maxima in aqueous solution, excellent

agreement is found with that predicted by the trend outlined in (1).

In reference 31, a log-log plot of specific adsorption versus ionic radius was used to support a

model of ionic hydration/contact adsorption to explain the extent of ion adsorption at the pzc

on mercury electrodes:

7

Figure 3: A log-log plot of specific adsorption versus ionic radius. Data reproduced from

reference 31.

However, for selected ions for which the polarizability is known, a plot of polarizability

versus specific adsorption (not log-log as above) gives an excellent linear relationship (figure

4):

8

Figure 4: Plot of polarizability at a mercury electrode versus specific adsorption for the halide

ions. Polarisability data taken from reference 27.

Interestingly, in reference 31, where the values of specific adption (S) for the halides

were obtained, the rather high values of S quoted for the alkyl ammonium cations did not

seem to correspond to S predicted on the basis of the uv data (methyl, ethyl and propyl

ammonium cations all give rise to νmax values above 53,000 cm-1 which, from Figure 2, would

indicate negligible cation adsorption at the PZC of mercury). However, it should be stressed

that the data in Figure 2 relate to 0.1 M aqueous solutions of ions in the absence of a

strongly adsorbing counter ion. In reference 31, the ionic surface excesses of the alkyl

ammonium ions at zero metal charge were derived from measurements using the iodide salts.

Frumkin demonstrated that the kinetics of many electrochemical reactions are influenced by

the presence of strongly adsorbed ions contained in the double layer [34]. The effects on the

reaction rate of having two strongly adsorbing ionic species was explained using a model of

"ion bridging pairs" formed on the electrode surface in which the distance between the

reacting anion and the cation which is its next nearest neighbour must be taken into account.

That is, enhanced adsorption of cations and anions at the PZC is to be expected if both ions

can specifically adsorb. Therefore, all uv data reported in the present work relate to fluoride

9

solutions of the alkyl ammonium cation where such effects as ion pairing at the metal

electrode are expected to be minimal. Later, it will be speculated why mixed anion and cation

layers should be more stable, so long as both ions specifically adsorb. The relevance of this

phenomenon to underpotential deposition will also be demonstrated.

In summary, so long as one knows the νmax for a particular ion (or better, the local

polarizability), it is possible to use Figures 2 and 4 to predict quantitatively the surface excess

of the ion at the PZC of the mercury electrode for a 0.1 M solution not containing strongly

adsorbed counter ions. In reference 35, at 0.1 M, the surface ionic excess of ethyl ammonium

cations at the PZC of mercury is reported and is almost zero as predicted (using ν max for this

ion) but in contradiction to the value of S reported for the iodide salt.

The majority of the uv-absorption maxima in aqueous solution reported in Figure 2

were of the charge transfer to solvent (CTTS) type [32]. It was interesting to note that, when

those ions which give rise to either n or some other intramolecular type of transition

were considered, without exception such ions decomposed into smaller molecular fragments

when in contact with an electrode. So, for example, perchlorate anions generate a uv-

absorption peak "at > 55.5 x 103 cm-1'' [36] which has been assigned as an intramolecular

transition. Perchlorate dissociates on many transition metal surfaces to give ultimately

adsorbed chloride ions [37 - 39]. Similarly, NO3−

, NO2−

, CO32−

and IO4−

give rise to n

or intramolecular transitions and all dissociate into molecular fragments when in contact

with transition metal electrodes [40 - 41]. However, other molecular ions such as bisulphate,

dihydrogen phosphate and acetate, which only generate CTTS-type uv-absorption peaks are

known to be quite stable at electrode surfaces.

Finally, before leaving Figure 2 entirely, it is worth commenting on the fact that the curve

levels off at approximately 65,000 cm-1 indicating that any ion giving rise to a uv-absorption

maxima at values in this range should display extremely weak specific adsorption. Hence, the

relatively poor extent of specific adsorption displayed by alkali and alkali earth cations may

now be readily explained on the basis of the very large energy required for a np6 → (n + l)s

transition [42] and low polarisabilities. In addition, according to figure 4, any ion with a local

polarizability of less than 2.8 Å3 will not specifically adsorb on Hg under the conditions of

reference 31. It is emphasised here that it is not being claimed that excited states (as would be

produced by an incident photon) are being generated at an electrode, rather the influence of

10

the electric field polarises the ion such that contributions to the total wave function describing

the electron distribution on the ion are weighted in favour of (unoccupied) excited states

relative to the electronic distribution encountered in the absence of a strong electric field (this

would account also for the strong correlation of anion stability with the nature of the excited

states mentioned earlier). This point will be elaborated later in section 3.2.3. Hence, it is

predicted that anions that undergo CTTS uv excitations (e.g. acetate, sulphate) will not

dissociate at electrode surfaces whereas those that exhibit more repulsive/anti-bonding

excited states (nitrate, perchlorate, carbonate) will tend to undergo molecular fragmentation.

3.1.2 The metal surface

Since localisation of charge is the overriding concern of this paper, what should one

expect of a given ion adsorbing on a particular metal? In terms of localisation of charge, d-

band metals will clearly be better than sp metals (sp bands display significant dispersion

whereas d- bands are relatively "flat"). For both the localisation of positive charge (anion

adsorption) and negative charge (cation adsorption) scaling of specific adsorption with the d-

band occupancy of electrons or holes respectively is expected. Therefore, the strength of

anion adsorption is predicted to follow the trend (Table 1):-

Hg, Bi, Ga, Pb < Ag, Cu, Au < Pd ≤ Pt < Rh < Ir < Ru (anion adsorption)

- localisation of positive charge

Hg, Bi, Ga, Pb < Ag, Cu, Au < Ru < Ir≤ Rh < Pt ≤ Pd (cation adsorption)

- localisation of negative charge

Indeed, many studies, including for example those by Frumkin and co-workers utilising

thermodynamic relationships [2], Wieckowski et al., by means of radio-tracer measurements

[43] and Attard et al. using chemical probes of local potentials of zero total charge (PZTC)

[44] have in general verified these trends at least as far as the transition metals are concerned,

although the positions of Pd and Pt should be reversed. However, it has been reported

previously [45] that, for a given anion (iodide), the strength of specific adsorption on the sp

metals is:

Hg > Bi > Pb > Cd > Ga …. (2).

11

Although the strong effect of water desorption has been proposed as the overriding

factor in determining this trend in ionic adsorbability [45], could charge localisation also

account for the experimental findings? Table 1 outlines the band filling and d-band width of

both sp and transition metal surfaces [46]. For transition metals in the same group relatively

little variation in the d-band electronic population is exhibited. However, as one descends the

group, the width of the d-band varies. So, for example, the d-band populations of Ni, Pd and

Pt are 8.97, 8.96 and 8.74 respectively. However, Ni gives rise to a rather narrow band width

in relation to Pd and Pt. That is, the d-band electrons are very much more localised for nickel

and, within the present model, nickel should give rise to extremely strong anionic adsorption.

So strong, in fact, that adsorbed water may also be polarised such that an irreversibly

adsorbed oxide is formed. In a similar vein, although palladium possesses a marginally higher

electronic population than platinum and, therefore, might be expected to adsorb anions less

strongly, the narrower d-band of Pd compared to Pt would lead to a greater propensity to

localise charge. Hence, it is proposed that a balance between electron population in a band

and the ability of the band to localise charge (band width) should ultimately govern the

“adsorbability” of a particular ion. In fact, the quantity (10 – d-band occupancy)/d-band

width (see Table 1) gives a rather good descriptor of anion adsorption strength and also water

interactions for Pt, Rh, Pd, Ir etc as also concluded in a slightly different manner by Norskov

et al [47]. Continuing this theme for the sp metals as a group [48 - 49], by taking the ratio of

p-electron density to s-electron density in the sp-band, it is possible to reproduce the trend

predicted in (2), at least for Ga < Cd < Pb < Bi (see Table 2). (The greater the proportion of p

character, the greater is the ability of the metal to localise charge). It must be remembered

that the large jump in p/s charge in going from lead to bismuth should be tempered by the

greater filling of the sp band for bismuth resulting in very low concentrations of charge

carriers [49]. A prediction made using the present model (Table 2), on the basis of p and s

contributions to the sp band, would be that for a given ion under identical conditions of

concentration and pH, the trend in metal-ion interaction would be:

Ga < Zn < In < Tl < Cd < Pb < Bi

Mercury was found not to follow such a trend in that, of the available band structure

calculations [50], it seems that for the liquid phase at STP, the occupied sp band is typically

70-80% s-character. However, Hg differs from all other sp metals in that its filled d-band cuts

across the bottom of the sp band leading to significant d-sp mixing [51]. This has been

confirmed by photoemission [52]. Hence, although the ratio of s to p character in the sp band

12

of mercury would suggest weak anion adsorption, we speculate that mixing of sp and d-band

components enhances the ability of Hg to localise charge such that, of the sp metals, it

displays the strongest anion adsorption. In summary, the ability to localise charge at a metal

surface may be thought of as a measure of the polarisability of the metal in an analogous

fashion to the isolated ion and may even relate to the magnitude of Friedel oscillations

generated for a given adsorbate, although at present, this point remains a speculation.

However, there are a number examples in the literature of surface structure determining the

amplitude of a Friedel oscillation whereby close-packed surfaces appear to generate Friedel

oscillation amplitudes somewhat greater than their more open counterparts [53 - 54]. Again,

the mechanism of differential strengths of anion adsorption at different surface sites may

therefore possibly be explicable in terms of this phenomenon.

For example, it is now well-documented that, for a given metal and ion, the strength

of specific adsorption depends upon the coordination of the atoms (for example in inhibiting

electrosorption of oxide) in the surface of the electrode, i.e. for an f.c.c. metal, the order of

anion adsorption strength is:

{111} > {100} > {110} [55]

It is well understood from surface physics that reduction in the coordination of surface

atoms in pure metals will result in an excess electron surface charge. The extent of this

charge redistribution being greater the more open is the surface. An experimental

manifestation of this effect is the initial state contribution to the core level shift observed by

X-ray photoemission [56]. Hence, the electronic population of the surface d-band is expected

to be greater than that of the bulk. Since the order of surface coordination for f.c.c. metals is:

{110} < {100} < {111}

the order of surface d-band filling should be:

{110} > {100} > {111}

But it has been postulated earlier that the extent of anion adsorption should scale with

electron population of the band. If this is correct, then for a given f.c.c. metal the {111}

close-packed surface (because of its smaller electron population relative to all other surface

geometries), should display the greatest amount of anion adsorption in accordance with

experimental findings. This is a quite general finding - a close-packed surface of a particular

metal should display greater anion adsorption than a more open surface since such a surface

is relatively the most depleted of electronic charge (c.f. the Friedel oscillation amplitudes

being greatest for close-packed surfaces [53 - 54]). Confirmation of this prediction as far as

cations are concerned may be gleaned from experimental observations of Zn2+ UPD on single

13

crystal platinum electrodes [57]. Minor changes in the hydrogen adsorption region are

observed for Pt{111} and Pt{100} in the presence of 10 -4 M ZnSO4 whereas, for Pt{110},

significant UPD of Zn2+ ions is reported. As expected, the enhanced electronic charge density

of the open {110} face in relation to the more close-packed orientations would give rise to

enhanced cation adsorption at a fixed electrode potential. Observation of UPD may indeed

simply reflect the larger magnitude of the Friedel oscillations generated by the zinc cation

adsorption on Pt{110} compared to Pt{111}.

3.1.3 Global and local charge and specific adsorption

Drawing all of these threads together - the local polarisability of the anion, the local

polarisability of the metal and also the strength of the solvent dipole/polarisability towards

stabilising the displaced charge - it is now possible to rationalise both specific adsorption or

"superequivalent" adsorption [58] (the notion of negatively charged ions adsorbing on

ostensibly negatively charged electrode surfaces) within a simple framework.

Figure 5 shows schematically the relationship between the shift in the PZC of a metal

in three different situations: (a) in the absence of specific adsorption (E2); (b) in the presence

of chloride (E3) and (c) in the presence of iodide (E4).

(a)

14

(b)

(c)

Figure 5 Local values of PZC for anions shifting to more negative potentials as

polarisability of the anion increases. The possibility of water adsorption sites ‘close to’ and

‘away from’ the anion co-existing on the surface is highlighted. (a) No specific adsorption of

anions, (b) specific adsorption of chloride, (c) specific adsorption of chloride and iodide.

It should be noted that the entire charge distribution of the system involving solvent

dipoles, metal polarisability and anion polarisability constitutes the specific adsorption bond

15

(all three are concerted processes) with E1 > E2 > E3 > E4. Clearly the shift in the PZC to

negative potentials caused by the presence of chloride (E2-E3) and that caused by the presence

of iodide (E2-E4) (with (E2-E3) > (E2-E4)) reflects explicitly the amount of local polarisation.

Ideas such as these have already been expressed by Ross in relation to stepped surfaces [59],

although the localisation of displaced charge by H2O molecules (or cations) was not

mentioned by this author. Nonetheless, in the opinion of the present author, the notion of a

local potential of zero charge proposed previously by Ross is absolutely correct and has been

explored in detail in another paper [44].

Capacitance and electrocapillarity measurements are macroscopic probes of electron

distribution and their limitation is that they cannot sense local values of PZC. Thus, just as

one may define the charge at a macroscopic surface in the following way [60]:

Q = Qsθs all sites s

where Qs is the local charge and θs is the coverage of sites of a particular type (steps,

terraces, point defects) so too may one do the same, for a {111} f.c.c. single crystal (all sites

equivalent), and assign sites as being at, near to, or far away from, an adsorbed anion. The

net excess charge in this system is always measured using capillarity or capacitance

experiments. However, if one uses chemical probes of local PZC, one may distinguish readily

between the differently charged sites [61]. In reference 61, data for submonolayer coverages

of palladium on Pt{111} being probed by N2O are discussed in relation to CO displacement

measurements of the same surfaces. The N2O reduction current maxima correspond to the

local PZTC of clean Pt{111} sites (0.34V vs. Pd/H) and pseudomorphic Pd{111} islands

(0.1V vs. Pd/H). This means that above 0.34 V the surface is positively charged at all anion

adsorption sites. However, between 0.1 V and 0.34 V, although the Pt{111} anion adsorption

sites are negatively charged (E < E Q=0Pt (111)

) the palladium island anion sites remain

positively charged (E > EQ=0Pd

). Only below 0.1V does one find that the whole surface is

negatively charged.

This is in accordance with the fact that the work function of Pt > Pd [62 - 63]. If it were

possible to measure the capacitance or electrocapillarity displayed by Pd on Pt{111} it should

be found that the PZTC measured would be a global average and would lie between 0.1 V

16

and 0.34 V, i.e. when the negative charge on the Pt islands was exactly balanced by the

positive and negative charges on the palladium. This indeed is found to be the case as

determined using CO charge displacement [61] and this idea is illustrated schematically in

figure 6.

Figure 6. The electrode as a ‘patchwork’ of locally charged states for Pd islands deposited on

Pt{111}[61]. The local pzc of the Pd and Pt areas are indicated by the vertical dotted lines.

The arrows indicate the flow of current associated with anion adsorption/desorption at Pt and

Pd sites.

Hence, one may now interpret the determination of PZC by capacitance and

electrocapillarity in a slightly different way. Figure 7 illustrates that it is possible for an

electrode to carry a net excess negative charge although locally, for example at the site of

anion adsorption, the metal may still actually be positively charged.

17

Figure 7 Interpretation of PZC as both a macroscopic and local effect by measurement

of surface charge and surface excess of anions (notated on the y-axis as ΓI-)

It is not until sufficiently negative potentials have been reached necessary to

"neutralise" the local excess positive charge at the anion site that complete desorption of

anions will be observed. Hence, we highlight once again that the unusual idea of negative

charges adsorbing on negatively charged surfaces (even if the electrolyte responds by

balancing the charge and maintaining electroneutrality [64]) need not be invoked to explain

specific or “superequivalent” adsorption. An alternative interpretation of "superequivalent"

adsorption would simply be that, although the total excess charge on an electrode may be less

than zero, locally if anions are still adsorbed, there is an excess local positive charge at the

anion adsorption site. As a corollary to this, the second element to the theory may now be

stated:

3.2 The Postulates of the Theory

The following postulates are listed to interpret specific adsorption:

(i) Anions will only adsorb on a locally positively charged site, i.e. when E ≥ EQ=0local

(ii) Cations will only adsorb on a locally negatively charged site, i.e. when E ≤ EQ=0local

18

A third postulate which has not as yet been discussed may be stated as follows:-

(iii) The three stages of ionic adsorption at electrode surfaces are:-

non-specific adsorption specific adsorption chemisorption.

In what follows, the criteria for deciding whether or not an ion is specifically

adsorbed will be described and the transformations necessary to facilitate passage from one

adsorption state into another will be discussed. A strong analogy will be drawn with the idea

of "precursor" states in gas/solid interactions [65].

From postulate (i), it is clear that as long as an anion "sees" a negatively charged

surface, it cannot specifically adsorb. This will be the working definition of "non-specific"

adsorption. To illustrate how this idea works, comparison of HSO 4−

specific adsorption at a

Pt electrode and the case of specific adsorption being absent is discussed. At pH = 1, the

PZTC of Pt/HSO4−

lies in the hydrogen adsorption region [2] (region A in Figure 8).

Figure 8 Schematic diagram comparing the cases of specific and non-specific anion

adsorption on the CV of polycrystalline Pt. For the case of specific adsorption, a mixture of

19

sites is highlighted, namely those local to the specifically adsorbed anion and those that are

not.

These potentials are associated with when anions commence adsorbing. As the

electrode potential is made positive of the PZTC, one reaches the double layer region (region

B). We ascribe this region to simultaneous adsorption of anions and re-orientation of water

molecules. If no anions were adsorbing, as E exceeded EQ=0local

, the water molecules would

eventually change their configuration from mainly hydrogen atoms bonded to electrode (E <

EQ=0local

(region A)) to mainly oxygen atoms bonded to the surface (E >EQ=0local

(region B)).

However, as discussed previously, the action of the co-adsorbed anions which specifically

adsorb is to cause (locally) reorientation of water molecules from oxygen down to hydrogen

down in order to stabilise displaced negative charge (enhance Friedel oscillation amplitude),

even though E > EQ=0. Hence, a mixture of water molecules in different bonding

configurations (“close to” and “away from” the adsorbed anion) are predicted to co-exist in

the double layer region. Furthermore, because polarised molecules are being discussed, it

should be noted that if both chloride and sulphate were present, the strength of adsorption of

the HSO4−

and C1- ions will increase as potential increases but to different extents due to

the difference in polarizability of the ions. The on-set of oxide formation corresponds to

the local PZC of the oxide surface [2] and is associated with the fact that some of the water

molecules adsorbed at the surface are highly polarised with their oxygen atoms bonded to the

surface (excess local surface positive charge for H2O molecules bonded to sites not in the

vicinity of the anion). We suggest that at sufficiently positive potentials, the water molecule

dissociates to form a proton which is desorbed from the double layer and a specifically

adsorbed hydroxide ion. However, if the electrode potential is made slightly more positive of

the on-set of oxygen adsorption, an irreversible electron transfer from the specifically

adsorbed hydroxide to the metal takes place with the formation of both surface and place-

exchanged Pt-OH species (i.e. the electron has shifted from the hydroxide anion onto the

metal). It has been demonstrated by Frumkin that the electrode side of the interface bears a

net negative charge after electrosorbed oxide is formed in accordance with this assertion [2].

Hence, although the electrode potential is increasing, the surface itself is negatively charged.

As more and more place-exchanged Pt-OH is formed, the adsorbed HSO 4−

(or chloride)

anions must desorb by postulate (i) (the surface increasing bares a negative charge) - this

20

phenomenon has actually been observed experimentally using in situ radio tracer methods

[66]. The whole adsorption process is summarised in Figure 8.

It is reported [67] also that Cl- delays the on-set of oxide formation relative to HSO 4−

on platinum and this may readily be rationalised within the present model by the fact that one

needs to be able to re-orientate many more H2O molecules from a "hydrogen down" to an

"oxygen down" configuration when Cl- is adsorbed compared to HSO4−

, to produce a given

amount of adsorbed oxide. This causes the on-set of oxide formation to shift to more positive

potentials. To obtain the amount of polarisation necessary to make H2O molecules generate

an oxide phase, a more positive potential is required to dissociate sufficient H2O molecules

(remember adjacent to the chloride, water is actually polarised by a negative charge excess of

the Friedel oscillation). When the extent of anion polarisation is larger, it can lead to very

large overpotentials for oxide adsorption. So, for example, on Pt {111} in 0.5 M H2SO4, the

on-set of oxide adsorption is delayed until 1.2V (RHE) [28]. Evidently, the combination of a

close-packed transition metal surface and a strongly adsorbing anion (associated with larger

Friedel oscillation and water in the ‘disfavoured’ orientation) inhibits oxide formation.

If one now applies similar arguments to a fluoride ion, it is expected to be negligibly

adsorbed in the double layer on the basis of its extremely large HOMO LUMO energy

gap / low polarisability. Nonetheless, given enough polarisation (high enough positive

electric field) or a substrate such as Pd, Ir or Rh which can localise positive charge more

readily than platinum, it too will specifically adsorb. In the Supplementary Information

section we show a CV of fluoride specifically adsorbing on a Pd adlayer on Pt{111}.

The difficulty for specific adsorption of F- on Pt is that, when the potential is

sufficiently positive to cause significant F- adsorption, the polarised H2O molecules referred

to earlier generate an oxide layer which, of course, leads to a negatively charged surface i.e.

as far as fluoride ions are concerned, their specific adsorption will remain negligible (in the

double layer region due to low intrinsic polarisability) or non-existent (in the oxide region

because the oxide bears an excess negative charge) i.e. fluoride is essentially non-specifically

adsorbed at all potentials. The same argument (in reverse) holds for cations such as Na+ in

that, when the electrode is sufficiently negatively charged to cause polarisation of the cation,

the solvent (H2O) is generating H2 gas since the HOMO LUMO gap on H2O is relatively

much smaller [25 - 26] and its polarisability exceeds that of the alkali metal cation, even

though a metal such as Pt should be very good at localising excess negative charge. However,

21

recent work using Pt{110}(highest propensity to localise negative charge of the three basal

planes) has demonstrated weak specific adsorption of sodium cations under some

experimental conditions [68]).

Therefore, the strength of specific adsorption for a particular anion may be thought of

as a continuous polarisation of the anion/metal/solvent system extending over a range of

potential from the local PZC of the clean surface to the on-set of oxide formation. Unlike for

metal-vacuum interfaces, this means that it is unlikely that a single, adsorption strength could

be assigned to species adsorbing at the electrochemical double layer, rather a gradation in

values as a function of potential should be expected.

To summarise, if either the local polarisability of the metal electrode is small or the

HOMO LUMO energy is large/polarizability small (or both!) an anion will exhibit very

little specific adsorption although even the most "non-specifically" adsorbed anion should

display some adsorption so long as E > EQ=0local

. This process will cease when the surface

bears a net negative charge such as when an oxide layer or a hydrogen monolayer is formed

[2].

3.2.1 The role of pH in specific adsorption

The strength of specific adsorption is actually pH dependent. That is, at low pH, a

chloride anion will strongly specifically adsorb on a platinum electrode for example whereas,

at high pH, this anion is essentially non-specifically adsorbed [69]. Even iodide gives rise to

such behaviour. Irreversible adsorption of iodide ions from an acidic solution onto a Pt{111}

electrode will block almost all hydrogen underpotential deposition (Hupd) since a strongly

chemisorbed layer of iodine is formed [70] (see later). However, if the same experiment is

performed at high pH, hydrogen underpotential deposition proceeds irrespective of iodide

ions being present in the solution [71]. Furthermore, the voltammogram reported for this

system is rather similar to that of Pt{111} in sulphuric acid in the region of initial iodide

adsorption [15] - anomalous spike included! The similarity is even more striking if one

compares the voltammetry of Pt{111} in 0.1M H2SO4 and bromide adsorption on Pt{111} at

pH = 10 [71].

These experimental findings pose two questions:

22

1 If ionic hydration is a critical factor in determining the strength of specific adsorption,

how shall this be affected by pH?

2 Why should a "universality" in the voltammetry of Pt{111} with regard to anion

adsorption be found which is a function of pH?

If we take question (1) first of all, in the opinion of the author, ionic hydration is not affected

by pH and so cannot be a major factor in determining the strength of specific adsorption.

Other experimental data, particularly radio-tracer measurements of anion adsorption and

exchange with the solution [72 - 73] have demonstrated that solvation is a dynamic process.

Water molecules are attached and detached from the central ion with corresponding rate

constants. On the electrochemical time scale, the collisions of the anion-water complexes

with the electrode are frequent and the probability that there will be a direct contact of the

"naked" or "partially naked" anion with the metal is high irrespective of charge density on the

ion. Therefore, can local polarisation explain these phenomena? Evidently the absolute

energy of the HOMO and LUMO levels of a halide should be independent of pH. Hence, the

local polarisability of the anion should remain the same. Let us suppose that we can

deconvolute the two independent processes that contribute to the specific adsorption of an ion

(S) discussed so far and assign them a value that we shall call f:

f 1 = local polarizability of anion

f 2 = local polarizability of metal

If this is indeed correct, then the specific adsorption strength S may be expressed as:

specific adsorption = S = f1(local polarisability) x f2(local polarisability) strength in aqueous solution of anion of metal.

……….(2)

If the "butterfly" peak voltammetry displayed by I- ions at pH = 13 behaves in the same way

as HSO4−

ions at pH = 1 on Pt{111}, because the butterfly peak is a measure of S, it follows

that:

S(I-, pH = 13) = S(HSO4−

, pH = 1) ...…… (3)

But from (2), we can substitute for S:

23

f1(I-) x f2(Pt{111}, pH = 13) = fl(HSO4−

) x f2(Pt{111}, pH = 1)

But fl(I-) > f1(HSO4−

)!

In order to maintain equality between S in both cases, the only conclusion that can be drawn

is that:

f2(Pt{111}, pH = 13) < f2(Pt{111}, pH = 1) …….(4)

What (4) represents within the model is that the local polarisibility of Pt{111} is a function of

pH. No experimental data or theoretical calculation regarding this point exists. However, if

local polarisability of the metal is linked to d-band occupancy (see earlier), then the model

predicts that there is a relative decrease in the surface electronic population of the d-band on

Pt{111} as pH decreases (resulting in an increased ability to localise excess positive charge

on the metal side of the double layer). This would manifest itself as Cl - (aq) ions being more

strongly adsorbed in acidic aqueous media whereas less adsorption would be found when

alkaline conditions prevailed. For I- in acid, chemisorbed iodide layers are formed whereas at

high pH, strong specific adsorption is observed [71, 74]. It is interesting to note that from the

work of Frumkin and Petrii, both polycrystalline Pt and Rh in 0.1 M KC1 under alkaline

conditions carry a negative surface free charge at all potentials [1]. It is predicted that, if 0.1

M KI was used as the background electrolyte in these experiments instead of KC1, for Pt at

least, a positive surface free charge under certain values of electrode potential would be

obtained even at high pH. A summary of how the surface excess of an anion should vary with

pH on Pt is given in Figure 9 and also its relation to the voltammetry of Pt.

24

Figure 9 Schematic representation of changes in the extent of chloride adsorption (Γ)

as a function of pH on polycrystalline platinum. PZC = local pzc of chloride site in figure 5.

To conclude this section, the "universality" of the voltammetry of Pt{111} may be

explained on the basis of transformations between non-specifically and specifically adsorbed

states, the degree of specific adsorption being a function of both f l and f2. The prediction

being that, if for a particular metal-anion pair the product of fl and f2 is similar, a similar

voltammogram should be obtained. The voltammetry under acidic conditions of Ir{111} [38]

and Rh{111} [43] as should recent work on bimetallic PtPd{111} and PtAg{111} surfaces

and chloride adsorption on Pt{111} [74 - 76] should be consulted regarding this point. In the

Supplementary Information section, our own data on sulphate adsorption on PtPd{111}

alloys and chloride specific adsorption on Pt{111} are compared showing almost identical

voltammetric behaviour. That is, by tuning the electronic structure of platinum to be more

like palladium (narrowing of d-band), sulphate behaves like chloride on Pt{111}.

3.2.3 Description of transformation between specific and chemisorbed state of

ions

The remaining question which needs to be answered is how an anion passes from a

specifically adsorbed state into a chemisorbed state? The classic example of this behaviour is

25

iodide adsorption on Pt [71]. In particular, reference should be made to the work of Hubbard

[69] and Abruna [74] in this regard. In order to explain chemisorption in the electrochemical

context, it is necessary to invoke ideas first proposed in relation to adsorption at solid

surfaces from the gas phase and then to modify these ideas slightly in the light of new

calculations as to the effect of strong electric fields on the HOMO LUMO energy on a

particular anion.

The physics of charge neutralisation [78], particularly in relation to ion scattering [79]

is now well developed. When an ion approaches a surface its energy levels may shift and

broaden due to interactions with the electron bands of the solid. This interaction can be

composed of at least three components i.e. Van der Waals interactions, overlap of electronic

wave functions (covalency) and a Coulombic image force of attraction. Although Hagstrum

[80] has shown that the polarisability of a cation is so small that the Van der Waals forces are

negligible for the range of energies encountered in ion scattering spectroscopy [81], this is

not the case when an anion or a polarisable cation approaches an electrode surface.

According to the present proposal, the specifically adsorbed state is a manifestation of

dispersion forces which are accentuated by the high degree of charge localisation engendered

by the presence of polar solvent molecules. Nonetheless, the formation of a "surface

molecule" between an incoming ion and the surface is now accepted as an essential element

of charge neutralisation mechanisms including promotion of electrons into 'molecular

orbitals' generated as a consequence of ion surface collision. Therefore, the relative energy

separation between electrons at the Fermi level (f) and an electron energy level on the ion

() determines the ease with which resonant neutralisation can proceed, i.e. if |f - o| is

"small", equilibration of charge between ion and surface is favourable [79]. If the hole state

on a cation or neutral, for example, lies well below the valence band, resonance neutralisation

is not possible since electron tunnelling through a large barrier is less probable than between

two shallow levels. Since electrons with energy e below the vacuum level, i.e. at the Fermi

level have the smallest barrier to penetrate, these are the electrons involved in charge

neutralisation by resonant exchange. Furthermore, if level broadening () is greater than |f -

o|, some overlap with f may occur such that charge transfer can proceed giving rise to

covalency in the ion-surface bond. Before trying to describe schematically why specifically

adsorbed states ("occupied" LUMO levels) may be stable relative to complete charge transfer

from ion to metal, some ab initio calculations describing the effect of a strong electric field

on the polarisability of a chloride ion in the gas phase are presented [82]. This is because

quantum calculations of ions approaching a surface in the gas phase tend not to include the

26

effect of strong electric fields as would happen for an anion approaching an electrode surface.

From Table 2 it is evident that a downshift in both the HOMO and LUMO level of C1- occurs

depending on the strength of the electric field. Of greater significance however is the fact that

the LUMO level is downshifted in energy to a much greater extent than the HOMO level, i.e.

as the electric field strength is increased, the HOMO → LUMO energy gap decreases. In the

present model, this would represent enhanced contribution from states in the LUMO level

such that the strength of specific adsorption (local polarisability) should increase in

proportion to the electric field strength. This effect has been incorporated into figure 10 in

order to explain the behaviour of the energy levels appropriate to specific adsorption and how

transformation into a "chemisorbed" state arises. We define two important energy separations

relevant to this discussion:

= |f - o| for the partially filled electron state in the LUMO level,

and

= |f - o| for the partially filled hole state in the HOMO level.

27

Figure 10 Changes in electronic energy states on an anion as a function of distance in the

electrical double layer. Distance in this case also corresponds to a potential dependent change

in the charged state of the electrode. For E < EQ = 0, the anion is non-specifically adsorbed

whereas for E > EQ = 0, the anion begins to specifically adsorb (postulate 1). Γ is the

broadening of a discrete energy level on the anion as a consequence of interaction with

electrons in the surface of the metal.

As the chloride ion approaches the electrode, there is a downshift and broadening in the

HOMO and LUMO energy levels. In addition, specific adsorption commences depending on

the separation in the HOMO and LUMO energies. Since the filled p6 HOMO level on the

aqueous ion is well-removed in energy from f in the aqueous phase [32, 33] and continually

downshifts in energy, as it approaches the electrode, the hole state generated as a

consequence of specific adsorption cannot decay via resonant exchange since is large.

Similarly, at distances far away from the electrode, or when the electric field strength is

small, the electron population of the LUMO level is also stable since is large. However, as

the LUMO level is strongly downshifted in energy due to both interaction with the electron

bands on the solid and the presence of a large electric field, the broadening of the LUMO

state becomes greater than |f - o|. This means that resonant neutralisation may occur with

depopulation of the LUMO level and electron transfer to the metal. At sufficiently large

values of , it may be seen that complete transfer of electronic charge from the ion to

themetal results.

In summary:-

E1 E2

C1-(aq) C1-

(aq) C1 + e

(non-specifically (specifically (chemisorbed) (on metal) adsorbed) adsorbed) E1 < E2

28

If a cation is now considered (figure 11), in contrast to anions, the LUMO levels are at

extremely high energies with respect to the vacuum level [42] such that their interaction with

the Fermi level electrons will be negligible i.e. is very large. However, is much closer in

energy to the Fermi energy of the electrons on the metal and is upshifted in energy as the

cation approaches the surface due to image potential effects [79].

Figure 11 Changes in electronic energy states on a cation as a function of distance in the

electrical double layer. Distance in this case also corresponds to a potential dependent change

in the charged state of the electrode. For E > Eσ = 0, the cation is non-specifically adsorbed

whereas for E < EQ = 0, the cation begins to specifically adsorb (postulate 2). Γ is the

broadening of a discrete energy level on the cation as a consequence of interaction with

electrons in the surface of the metal.

Thus, although, as for Cl-, polarisation of the cation results, dependent on the HOMO →

LUMO energy gap, as specific adsorption proceeds, the interaction of the partially filled hole

29

state on the HOMO level with the Fermi electrons dominates the charge neutralisation

process ( > | f - o|). The net reaction taking place is thus:

E1 E2

Cs(aq )+

Cs+ Cs + hole(non specifically (specifically (chemisorbed) (on metal) adsorbed) adsorbed) E2 < E1

If chemisorption is defined as when covalency between ion and metal begins, strictly,

this takes place when f-

However, if f- , charge neutralisation cannot occur even though polarisation

on the ion leading to promotion from HOMO to LUMO levels is efficient. That is, although

the notion of "partial electrovalency" as discussed by Lorenz [9, 10], Vetter and Schultz [83]

and others [84, 85] is almost certainly taking place when f - for f - no

electron transfer from the specifically adsorbed ion to the metal is expected. That is, specific

adsorption may be thought of as ranging (as a function of increasing potential) from a

polarised ion interacting with a polarised metal locally without any electron transfer to (as

electron transfer becomes more feasible) a depolarisation of both anion and metal following

electron transfer. However, according to this interpretation, the transfer of electrons to the

metal occurs via occupation of virtual states created as a result of ion adsorption at the

surface. Using the above description of "specific" and chemisorption behaviour, one may

now rationalise the reported dependence of PZC on two electronically different metals, Hg

and Pt [86] in the presence of iodide. Strong specific adsorption of iodide on Hg leads to a

negative shift in the PZC relative to the clean surface [45]. This is because in order to

"neutralise" the excess positive charge at the iodide specific adsorption site, a more negative

electrode potential is required such that desorption takes place when E<EQ=Olocal/I−

(postulate

1). In contrast, for iodide strongly chemisorbed on Pt, radio-tracer measurements of surface

excess quantities show that the PZC shifts to more positive potentials relative to the clean

surface [86]. As recognised by Frumkin [1], this means that the electron from the iodide must

reside on the metal side of the interface in this case such that the surface bears a negative

charge with respect to the electrolyte (like an oxide), the adsorbed iodine of course being

neutral and covalently bonded to the surface. The reason iodide is only specifically adsorbed

on Hg is due to the small value of f2 (Pt in contrast has a partially filled d-band so exhibits a

larger value of f2). Similarly, for the cation Tl+, although "normally", specifically adsorbed

30

cations give rise to a positive shift in PZC relative to the clean surface (see, for example, the

adsorption of alkyl ammonium cations) [1], on platinum, complete charge transfer from Pt to

T1+ takes place such that a metallic T1 overlayer is formed with an excess positive charge on

the Pt [86]. For Tl+ on Pt, a negative shift in PZC is reported in accordance with expectations

[1].

What has just been described is an example of the underpotential deposition of one

metal upon another [87]. Can the model shed some light on processes associated with the

electrochemical deposition of metals?

3.3 Underpotential deposition in terms of localised charges and polarisabilities

Certainly two points need to be stressed:

(A) According to postulate (ii), cations will only adsorb on a locally negatively charged

site, i.e. when E<EQ=Olocal

.

(B) From the previous discussion, specific adsorption of cations shifts EQ=0 to more

positive potentials whereas chemisorption ( f- ) tends to shift the PZC in the

opposite direction. Hence does one observe such behaviour in relation to underpotential

deposition?

According to the model, the strongest specific adsorption of anions will take place on

close-packed surfaces. Therefore, we take as an example copper UPD on Pt{111} since data

has already been published in relation to this system for many different types of anion and

also this surface should give rise to large effects [88 - 92]. Figure 12 shows schematically the

UPD stripping peak of copper on Pt{111} as a function of anion type at constant anion

concentration. The critical parameter to be discussed is "t", the separation between the two

UPD peaks.

31

(a)

(b)

Figure 12 Schematic representation of copper UPD in acidic aqueous media on

Pt{111} in terms of local charges for (a) sulphate and (b) chloride containing electrolyte.

Reference electrode is copper wire in contact with the electrolyte (10-3 M copper perchlorate

in 0.1 M sulphuric acid and 10-3 M copper perchlorate in 0.1 M sulphuric acid + 10-3 M NaCl

respectively). At potential E a continuous monolayer of copper is formed. At potential A, no

copper is adsorbed although interaction of copper cations with surface water associated with

32

the anion is indicated by the dashed line. The co-adsorbed copper-anion adlayer transforming

into a single copper monolayer via place exchange of the anion by copper is based on

reference 90.

At potential "A", if one compares the situation of C1- and HSO4−

:

EQ=0H 2O (Cℓ−)

> EQ=0

H 2O (HSO4− )

i.e. at the co-adsorbed H2O site, for a given potential, the H2O site associated with

specifically adsorbed C1- is much more negatively charged than for HSO4−

. This site will be

the preferential adsorption site for the aqueous copper cation. By postulate (ii), this means

that the onset of cation adsorption should take place at more positive potentials in the

presence of C1- as compared to HSO4−

. From Figure 12, this is indeed seen to be the case.

Cation adsorption will result in a co-adsorption layer in which a synergistic charge

delocalization will cause both anion and cation to be more strongly adsorbed (potential B).

Theoretical and experimental results [88 - 90, 93] confirm the presence of co-adsorbed anion

and cation layers on {111} surfaces at this stage of deposition. The present theory predicts

why these co-adsorbed layers should be stable. However, to displace the C1 - anion

completely, by postulate (i):

E < EQ=0Cℓ−

But clearly, based on polarisability:

EQ=0Cℓ−

< EQ=0HSO

4−

Hence t (Cl-) > t (HSO4−

)

i.e. the overpotential required to desorb the C1- from a co-adsorbed anion and cation layer is

much greater than that needed to displace HSO4−

. Thus t is (according to the model) a direct

measure of the local polarisation of the anion in the co-adsorbed adlayer and is largest for

chloride relative to bisulphate. However, why should t decrease as one changes to the more

polarisable Br- ion [88]? The answer lies in the fact that if on the LUMO of the Br- is > f-

then chemisorption commences, that is charge transfer from the bromide to the metal is

33

facilitated. Accordingly, this corresponds to a depolarisation of the co-adsorbed layer and a

decrease in t. In the limit of iodide adsorption, complete charge transfer from the iodide to the

metal means that t is negligibly small and only one UPD peak may be observed. This is also

the case for other strongly adsorbed anions [93]. However, if copper UPD was carried out at

higher pH, the model predicts that an increase in t will occur as charge is transferred back

from the metal to the iodide due to a decrease in the ability of Pt{111} to localise positive

charge on the anion adsorption site (see earlier). The analogy of this situation with "ion

bridging pairs" should also be noted [34].

3. Discussion of the present model in relation to previous theories of specific adsorption.

It was Grahame who originally postulated that ions contained in the inner layer are

specifically adsorbed [94]. He assumed that specific adsorption involved covalent bonding

between the adsorbed anion and the electrode. However, Levine et. al rejected this proposal

[95]. These workers suggested that the image energy was the origin of specific adsorption.

Calculations by Bockris indicated that the image force was too small to account for specific

adsorption [96]. Bockris also rejected the notion of covalent bonding as the primary

component of the specific adsorption bond [96]. Instead, he supposed that the degree and type

of ionic solvation were predominant in determining the strength of specific adsorption.

According to Bockris, the concept of covalency should be discounted on the grounds that, for

the halide series at least, their specific adsorption on mercury was the reverse of the Hg-

halide bond energy in the gas phase. Hence low hydration was seen as a necessary condition

for specific adsorption to occur. However, contradicting this view, Armstrong [97] reported

the very strong specific adsorption of the sulphide ion on mercury in spite of the fact the

sulphide is strongly hydrated. Also, measurements in aqueous solution have shown that the

weakly solvated PF6−

ion is only slightly adsorbed on mercury [98].

Hence, both in terms of the present theory and previous studies, the notion of

desolvation and "contact adsorption" being a necessary condition for specific adsorption to

take place seems incorrect. An article by Barclay [99] attempted to rationalise specific

adsorption in terms of soft and hard acids and bases (SAHB).

Barclay invoked Jorgensen's [100] view of "hardness" as being associated with "low

polarisability and isolated electronic ground states" whereas softness implied "high

polarisability and low-lying electronic states". The metal surface was assumed as having

34

properties of a "soft acid" (metals in zero valent states are soft). In essence the specific

adsorption bond was considered to be more akin to a "donor-acceptor" complex and soft-soft

interactions would lead to strong specific adsorption whereas hard-hard interactions (at

potentials far removed from the metal PZC) would generally be ionic in nature. Hence, for

the halides, the order of specific bond strength should not be explained in terms of the free

energy of covalent bond formation:

M(g) + 1/2X2(g) → MX(g) (a)

but rather as a donor-acceptor complex:

M+(g) + X-(g) → MX(g) (b)

The free energy changes derived from reaction (b) do follow the expected trends in specific

adsorption [99].

The parallels with the present discussions are obvious. Barclay's ideas of specific

adsorption in terms of a SAHB are essentially the same as those expressed here although at a

microscopic level, the importance of co-adsorbed water, the effect of the electric field, and

the relevance of electronic band structure was not discussed. Within the present context also,

"partial covalency" is possible under well defined conditions as described previously. For

ions whose energy levels do not straddle the Fermi energy of the metal electrons, the specific

adsorption state is associated with intra molecular reorientation of electrons on the anion

and on the metal separately.

Conclusion

A phenomenological theory of electrosorption has been presented based on notions of

local charged states on the electrode and local polarisability of the adsorbate. A number of

double layer processes may readily be rationalised within the model and quantitative

predictions of the extent of ion adsorption at the PZC of mercury electrodes may be made

based on the UV absorption of the ion in aqueous solution. The nature of the UV absorptions

(intramolecular of CTTS) exhibited by an aqueous ion may also be used to predict its

molecular stability in contact with an electrode. It is suggested that the model provides a

rational basis for understanding much electrosorption behaviour and, in a future paper, will

also be invoked to predict electrocatalytic trends and understanding of adsorption processes

associated with neutral molecules [101].

35

Acknowledgements

Fruitful discussions during the writing of this paper with A. Wieckowski, Juan Feliu,

Nenad Markovic, R. Parsons and T. Vandernoot are acknowledged. Thanks also to Jin-Chao

Dong for assistance in preparing the figures.

References

1. Comprehensive Treatise of Electrochemistry: Volume 1 eds. J.O.M. Bockris, B.

Conway and E. Yeager, (Plenum, New York, 1980).

2. A.N. Frumkin, N.A. Balashova and V.E. Kazarinov, J. Electrochem. Soc., 113 (1966)

1011.

3. D.C. Grahame and R.B. Whitney, J. Am. Chem. Soc., 64 (1942) 1548.

4. A. Frumkin, O. Petri, A. Kossaya, V. Entina and V. Topolev, J.

Electroanal. Chem., 16 (1968) 175.

5. S. Trasatti, Zeit. Physikal. Chem. Neue Folge, 98 (1975) 75.

6. J.O.M. Bockris and P.P.S. Saluja, J. Phys. Chem., 76 (1972) 2140.

7. H. Wroblowa, Z. Kovac and J.O.M. Bockris, Trans. Farad. Soc. 61, (1965) 1523.

8. Modern Electrochemistry, J.O.M. Bockris and A.K.N. Reddy, Plenum, New York,

1970.

9. W. Lorenz and G. Salie, Z. Phys. Chem. (Leipzig) 218 (1961) 259.

10. W. Lorenz, Phys. Chem., 224 (1963) 145.

11. D.J. Barclay, J. Electroanal. Chem., 28 (1970) 443.

12. R.W. Evans, G.A. Attard, J. Electroanal. Chem. 345 (1993) 337.

13. D.J. Watson and G.A. Attard, Electrochim. Acta, 46 (2001) 3157.

14. F.J. Vidal-Iglesias, A. Al-Akl, D. Watson and G.A. Attard, J. Electroanal. Chem.,

611 (2007) 117.

15. J. Clavilier, R. Faure, G. Guinet, R. Durand, J. Electroanal. Chem., 107 (1980) 205.

16. M.W. Finnis, Surf. Sci., 241, (1991) 61.

17. J. Villain, M. Lavagna, P. Bruno, Comptes Rendus Physique, 17 (2016) 276.

18. B. Chatterjee, K. Byczuk, J. Phys.: Conference Series 592 (2015) 012059.

19. P. Han, P. S. Weiss, Surf. Sci. Reports, 67 (2012) 19.

20. S. Kucharski, Y.S. Lee, G.D. Purvis and R. Bartlett, Phys. Rev. A, 29 (1984) 1619.

21. G. Dierksen and A. Sadlej, Chem. Phys. Letts., 84 (1981) 390.

22. E. Bichoutskaia  and N. C. Pyper, J. Phys. Chem. C, 111 (2007) 9548.23. P. Jungwirth and D. J. Tobias, J. Phys. Chem. A, 106 (2002) 379.

36

24. K. Fajans, Z. Phys. Chem. B, 24 (1934) 103.

25. K. Fajans, J. Phys. Chem., 74 (1970) 3407.

26. S. Yoo, Y. A. Lei, X. C. Zeng, J. Chem. Phys., 119 (2003) 6083.27. H. Coker, J. Phys. Chem., 80 (1976) 2084.

28. J. Clavilier, A. Rodes, K. E1 Achi and M.A. Zamakhari, J. Chim. Phys., 88 (1991)

1291.

29. P. Salvador,  J. E. Curtis, D. J. Tobias, P. Jungwirth, Phys. Chem. Chem. Phys., 5

(2003) 3752.

30. S.T. Howard, G.A. Attard and J.F. Lieberman, Chem. Phys. Letts., 238 (1995) 180.

31. T.N. Anderson and J.O.M. Bockris, Electrochim. Acta., 9 (1964) 347.

32. M.J. Blandamer and M.F. Fox, Chem. Rev., 70 (1970) 59.

33. G. Stein and A. Treinin, Trans. Faraday Soc., 55 (1959) 1091.

34. A.N. Frumkin, Trans. Faraday Soc., 55 (1959) 156.

35. G.J. Ogilby, Ph.D. Thesis, University of Bristol, 1966.

36. J. Barrett, Ph.D. Thesis, University of Manchester, 1959.

37. G. Horanyi and I. Bakos, J. Electroanal. Chem., 331 (1972) 727.

38. A. Ahmadi, E. Bracey, R.E. Evans and G. Attard, J. Electroanal. Chem., 350 (1993)

297.

39. C.K. Rhee, M. Wasberg, P. Zelenay and A. Wieckowski, Catalysis Letts., 10 (1991)

149.

40. O. Petrii and T. Safonova, J. Electroanal. Chem., 331 (1992) 897.

41. E. Morallon, J. Vazquez, A. Aldaz and J. Clavilier, J. Electroanal. Chem., 316 (1991)

263.

42. Charlotte Moore, Atomic Energy Levels, Circular of the National Bureau of

Standards, Circular 467, (1958), ed. A.V. Astin.

43. A. Wieckowski, P. Zelenay and K. Varga, J. Chim. Phys., 88 (1991) 1247.

44. G.A. Attard and A. Ahmadi, J. Electroanal. Chem. 389 (1995) 175.

45. S. Trasatti, J. Electroanal. Chem., 65 (1975) 815.

46. M. Brejnak and P. Modrak, Surf. Sci., 247 (1991) 215.

47. B. Hammer, J.K. Norskov, Surf. Sci., 343 (1995) 211.

48. D.A. Papaconstantopoulos, Handbook of the Band Structure of Elemental Solids,

(Plenum, New York, 1986).

49. X. Gonze, J.P. Michenaud and J.P. Vigneron, Phys. Rev. B., 41 (1990) 11 827.

50. L.F. Mattheiss and W.W. Warren, Phys. Rev. B, 16 (1977) 624.

37

51. S.C. Keeton and T.L. Loucks, Phys. Rev., 152 (1966) 548.

52. R.W. Joyner, M.W. Roberts and K. Yates, Surf. Sci., 87 (1979) 501.

53. J.M. Li, Jianjun Wang, Q. Sun, Yu Jia, Physica B, 406 (2011) 2767

54. Jia-Jun Tang, Xiao-Bao Yang, LiuZhang Ou Yang, Min Zhu, Yu-Jun Zhao, J. Phys.

D: Appl. Phys., 47 (2014) 115305.

55. M. Bamboa-Aldeco, E. Herrero, P. Zelenay and A.Wieckowski, J. Electroanal. Chem,

348 (1993) 451.

56. W.F. Egelhoff, Surf. Sci. Repts., 6 (1987) 253.

57. S. Taguchi, A. Aramata, M.A.Quaiyyum and M. Enyo, J. Electroanal. Chem., 374

(1994) 275.

58. A.N. Frumkin and A.S. Titievaskaya, Zh. Fiz. Khim., 31 (1957) 485.

59. P.N. Ross, J. Chem. Phys., 88 (1991) 1353.

60. G. Valette and A. Hamelin, J. Electroanal. Chem., 45 (1973) 301.

61. G.A. Attard, O. Hazzazi, P.B. Wells, V. Climent, E. Herrero, J.M. Feliu, J.

Electroanal. Chem., 568 (2004) 329.

62. R. Vanselow and X.Q.D.Li, Surf. Sci. Letts., 262 (1992) L200.

63. M. Methfessel, D. Hennig and M. Scheffler, Phys. Rev. B., 46 (1992) 4816.

64. J.O.M. Bockris and M.A. Habib, Z. Phys. Chem. Neue Folge, 98 (1973) 43.

65. H. Kang and W.H. Weinberg, Surf. Sci.,299/300, (1994) 755.

66. G. Horanyi and W. Wieckowski, J. Electroanal. Chem., 294 (1990) 267.

67. B.E. Conway, D.M. Kovait, J. Chem. Soc., Faraday Trans., 77 (1981) 2341.

68. G.A. Attard, K. Hunter, E. Wright, J. Sharman, R. Martinez-Hincapie, J.M. Feliu, J.

Electroanal. Chem., 793 (2017) 137.

69. F. Lu, G.Salaita, H. Baltruschat and A.T. Hubbard, J. Electroanal.Chem.,222 (1987)

305.

70. J. Clavilier, R. Albalat, R. Gomez, J.M. Orts and J.M. Feliu, J.Electroanal. Chem.,

360 (1993) 325.

71. E. Morallon, J. Vazquez and A. Aldaz, J. Electroanal. Chem., 334 (1992) 323.

72. P. Zelenay and A. Wieckowski, J. Electrochem. Soc., 139 (1992) 2552.

73. P. Zelenay, M. Gamboa-Aldeco, G. Horanyi and A. Wieckowski, J.Electroanal.

Chem., 357 (1993) 307.

74. J.H. White and H.D. Abruna, J. Phys. Chem., 92 (1988) 7131.

75. N. Garcia-Araez, V. Climent, E. Herrero, J. Feliu, J. Lipkowski, J. Electroanal.

Chem., 576 (2005) 33.

38

76. J. Clavilier, L.H. Klein, A. Vaskevich, A.A. El-Shafei, J. Chem. Soc. Farad. Trans.,

92 (1996) 3777.

77. T.J. Schmidt, N.M. Markovic, V.Stamenkovic, P.N. Ross, G.A. Attard, D.J. Watson,

Langmuir, 18 (2002) 6969.

78. D.P. Woodruff, Nucl. Inst. and Methods., 194 (1982) 639.

79. S.R. Kasi, H. Kang, C.S. Sass and J.W. Rabalais, Surf. Sci. Repts., 10 (1989) 1

80. H.D. Hagstrum, Phys. Rev., 96 (1954) 336.

81. W. Heiland and E, Taglauer, Nucl. Instr. and Methods, 132 (1976) 535.

82. S. Howard, (private communication). The basis sets used in these calculations were

the same as those in reference 21.

83. J.W.Schulze and K.J. Vetter, J. Electroanal. Chem., 44 (1973) 63.

84. R. Parsons, Advances in Electrochemistry and Electrochemical Engineering, eds:

C.W. Tobias and P. Delahay, Vol. 7, Interscience publishers (1970) New York.

85. B.B. Damaskin, Elektrokhimiya, 5 (1969) 771.

86. T. Ya. Kolotyrkin, O.A. Petrii and V.E. Kazarinov, Elektrokhimiya, 10 (1974) 1352.

87. D.M. Kolb, Advances in Electrochemistry and Electrochemical Engineering, 11

(1978) 125.

88. J.H. White, H.D. Abruna, J. Phys. Chem., 94 (1990) 894.

89. L. Blum and D.A. Huckaby, J. Electroanal. Chem., 375 (1994) 69.

90. R. Gomez, J. M. Feliu and H.D. Abruna, J. Phys. Chem.,98 (1994) 5514. 68.

91. N. Markovic, P.N. Ross, Langmuir, 9 (1993) 580.

92. M. S. Zei, K. Wu, M. Eiswirth, G. Ertl, Electrochim. Acta, 45 (1999) 809.

93. E. Herrero, L. J. Buller and H. D. Abruna, Chem. Rev., 101 (2001) 1897.

94. D.C. Grahame, Chem. Rev., 41 (1947) 441.

95. S. Levine, G.M. Bell and D. Calvert, Can. J. Chem., 40 (1962) 518.

96. J.O.M. Bockris, M.A.V. Devanathan and K. Muller, Proc. Roy. Soc., London, 274A

(1963) 55.

97. R.D. Armstrong, D.F. Porter and H.R. Thirsk, J. Electroanal. Chem., 16 (1968) 219.

98. E.R. Nightingale, Chemical Physics of Ionic Solutions, Wiley (1966) New York.

99. D.J. Barclay, J. Electroanal. Chem., 19 (1968) 318

100. C.F.K. Jorgensen, Structure and Bonding, 1, (1966) 234.

101. G.A. Attard (to be published).

39

Table 1

Metal d-bandpopulation

d-bandwidth/e V

(10 - d-bandpopulation)/ d-band width

Ni* 8.97 3.7 - - 0.278

Cu* 9.91 2.8 - - 0.032

Ru* 7.24 8.6 - - 0.321

Rh* 7.99 7.1 - - 0.283

Pd* 8.97 5.5 - - 0.187

Ag* 10.00 3.6 - - 0.000

Ir* 7.65 9.2 - - 0.255

Pt* 8.74 7.3 - - 0.173

Au* 9.89 5.3 - - 0.021

s-electronpopulationof sp-band

p-electronpopulationof sp-band

p/s ratioin sp-band

Ga** - - 2.19 0.65 0.2968

In** - - 2.03 0.92 0.4532

T1** - - 1.71 1.04 0.6080

Zn** - - 1.22 0.47 0.3852

Cd** - - 1.07 0.75 0.7009

Pb** - - 2.18 1.81 0.8303

Bi*** - - 1.99 3.01 1.512

* Data taken from Reference 46.

** Data taken from Reference 48.

*** Data taken from Reference 49 and integrated.

40

Table 2

Calculation of the Effect of Electric Field Strength on

HOMO and LUMO Levels of Gas Phase Chloride Ion*

Field/au HOMO/a.u. LUMO/a.u. GAP/cm-1

0 -0.1503 +0.0566 45411

0.001 -0.1503 +0.0544 44928

0.003 -0.1504 +0.0414 42097

0.005 -0.1506 +0.0230 38102

** 0.01 -0.1516 -0.0291 26887

** 0.02 -0.1564 -0.1385 3929

* Basis sets used in the calculation are identical to those in Reference 21.

** Fields of 0.001 a.u. and above are extremely large ( > 5 x 109 Vm-1)

41