Electrocatalysts Prepared by Galvanic Replacement

Galvanic replacement is the spontaneous replacement of surface layers of a metal, M, by a more noble metal, M noble , when the former is treated with a solution containing the latter in ionic form, according to the general replacement reaction: nM + mM noble n+ → nM m+ + mM noble. The reaction is driven by the difference in the equilibrium potential of the two metal/metal ion redox couples and, to avoid parasitic cathodic processes such as oxygen reduction and (in some cases) hydrogen evolution too, both oxygen levels and the pH must be optimized. The resulting bimetallic material can in principle have a M noble-rich shell and M-rich core (denoted as M noble (M)) leading to a possible decrease in noble metal loading and the modification of its properties by the underlying metal M. This paper reviews a number of bimetallic or ternary electrocatalytic materials prepared by galvanic replacement for fuel cell, electrolysis and electrosynthesis reactions. These include oxygen reduction, methanol, formic acid and ethanol oxidation, hydrogen evolution and oxidation, oxygen evolution, borohydride oxidation, and halide reduction. Methods for depositing the precursor metal M on the support material (electrodeposition, electroless deposition, photodeposition) as well as the various options for the support are also reviewed. higher standard potential) then, due to the difference in their standard electrochemical potentials, E 0 noble − E 0 > 0, and provided that the ionic form of M is stable under the given experimental conditions (of temperatue, pH, complexing agents etc.), the following reaction is thermodynamically favored and can take place spontaneously:


Thermodynamic Considerations
When a metal, M (e.g., Cu, Fe, Co, Ni, Al, etc.), is immersed in a solution containing ions of a more noble metal, M noble n+ (e.g., Pt, Au, Pd, Ag, Ru, Ir, Rh, Os, etc.-i.e., a metal with a higher standard potential) then, due to the difference in their standard electrochemical potentials, E 0 noble − E 0 > 0, and provided that the ionic form of M is stable under the given experimental conditions (of temperatue, pH, complexing agents etc.), the following reaction is thermodynamically favored and can take place spontaneously: This reaction, which bears similarities with transmetalation reactions between metal complexes [1] and is also known as immersion plating [2] in the plating industry and galvanic replacement [3] in materials chemistry, can be considered to originate from the coupling of the following two half-reactions: M noble n+ + ne − ↔ M noble (E 0 noble ) M m+ + me − ↔ M (E 0 ) (3) and results in the (complete or partial) dissolution of M as M m+ ions and the deposition of M noble in metallic form, either on surface sites of remaining M, or on top of already deposited M noble atoms, or on nearby substrate locations (provided that the latter is conducting).As examples (relevant to the preparation of electrocatalysts or optically active particles; see below) one may consider the galvanic replacement of metallic Cu by Pt or that of metallic Ni by Ag: 2Cu + PtCl 6 2− → Pt + 2Cu 2+ + 6Cl − (4) Alternatively, if for a given metal M and pH conditions the oxidized form of M can only be present as an oxide, M 2 O m , or hydroxide, M(OH) m , then the deposition of metallic M noble is not coupled with the release of M m+ in the solution but the formation or further growth of oxides/hydroxides of M, either at freshly reduced M sites or onto already present oxide layers.(Then, a more accurate description of the process should employ the term galvanic deposition-oxide growth instead of galvanic replacement and it is often referred to as spontaneous deposition.)Hence, in the case of oxide formation, galvanic deposition can be described by the following general reaction: the driving force now being the E 0 Mnoble − E 0 M 2 O m > 0 difference.As examples (relevant to the preparation of electrocatalysts or the deposition of metal ions on stainless steel containers; see below) one may consider the galvanic deposition of metallic Pt on freshly reduced Ti and Ru as well as that of Ag on Cr-containing materials: As mentioned above, an indication of the thermodynamic driving force for galvanic replacement/deposition is given by the difference in the standard potentials of the half-cell reactions involving the two metals; a series of such half-cell reactions often encountered in galvanic replacement/deposition and their standard potentials (taken from [4]) is given in Table 1.The left-hand side of the table contains metals that are usually deposited during the process (the forward of the half-cell reaction) whereas the right-hand side contains those that are usually dissolved/oxidized (the reverse of the half-cell reaction).The larger the difference between the potential of a left-hand side reaction and that of a right-hand side one, the more favorable their coupling into a galvanic replacement/deposition process.For example, from the reactions presented in Table 1 the two most favorable reactions would be the deposition of Au on freshly reduced Cr or Al with the simultaneous oxidation of the latter into Cr(OH) 3 or Al(OH) 3 , since the combination of the relevant half-cell reactions have a potential difference of +1.002 − (−1.126) = +2.128V vs. SHE (Standard Hydrogen Electrode) and +1.002 − (−2.300) = +3.302V vs. SHE respectively.It is obvious that before embarking on thermodynamic calculations that provide an estimate of the tendency of a galvanic replacement/deposition reaction to occur one should take into consideration the following: (i) the Pourbaix diagram of the metal M that is tested for replacement, to identify the forms in which it may exist under the given pH and aeration conditions and the corresponding standard potentials (for example, in aerated neutral solutions Cr should be present as Cr 2 O 3 [5] rendering galvanic deposition of Ag simply via reaction (9) impossible to take place on defect-free chrome surfaces); (ii) standard potentials should be corrected to their equilibrium values, taking into account the exact metal ions concentration [6]; and (iii) the possibility of metal complexation, depending on the presence of ligands, which should be taken into account when considering the half reactions since this may not only affect the corresponding potential but also the valence of the metals and the stoichiometry of the reaction (as pointed out by Brankovic and co-workers [6,7], who provided evidence that, in the absence of any other than Cl − ligands, reaction (4) above may be replaced by one involving the replacement of four Cu atoms by one Pt with the formation of the stable Cu(I) complex of CuCl 2 − ).Indeed, one has to stress that some of the reduction potentials tabulated in the right-hand columns of Table 1 may decrease significantly in the presence of strongly complexing ligands (see, for example, the case of the Cu(I)/Cu couple in the presence of Cl − ).

Kinetic Considerations
In estimating the rate of galvanic replacement/deposition in practice (which in turn may determine the morphology, structure and properties of the deposits) one has to look also into the kinetics of the process.There are rather few papers studying the time evolution of galvanic replacement processes.Most of them deal with the kinetics of the galvanic replacement of an underpotentially deposited (upd) monolayer of Cu or Pb by more noble metals.The first paper was by Dimitrov and co-workers [8] followed by the works of Brankovic and co-workers [9,10].These researchers have developed models whereby the coverage θ of the upd layer that is being replaced is correlated to time and, by the correlation of θ with the substrate potential via an adsorption isotherm, open circuit potential (ocp) vs. time expressions were obtained and validated by monitoring the substrate ocp as the galvanic replacement proceeds.They have treated cases of both mass transfer and kinetic control.More recent publications have taken into account the effects of solution composition [11] as well as those of surface heterogeneity and sequential multilayer formation [12].
When it comes to galvanic replacement of metal multilayers from nanoparticles or nanoparticulate films, there is a single paper that employed chronopotentiometry to monitor Pt deposition on Cu layers from acidic solutions; that process was found to be Pt-species diffusion limited, since both the kinetics of Pt deposition and of Cu dissolution in acid were very fast [13].Given the fact that galvanic replacement can be essentially viewed as a corrosion reaction whereby a metal substrate is oxidized/dissolved (anodic reaction) while a more noble metal is reduced/deposited (cathodic reaction), we hereby propose to treat galvanic replacement kinetics by analogy to corrosion kinetics, according to a phenomenological approach.Corrosion kinetics are characterized by a corrosion current density, j corr , which in turn can be related to a metal dissolution rate via Faraday's law.By analogy to j corr one can define (under no mass transfer limitations) for the combination of reactions similar to (2) and (3) above a galvanic current density, j galv , according to [14] as: where j 0,M noble and j 0,M are the exchange current densities of reactions similar to (2) and (3) respectively, β Mnoble , β M the Tafel slopes of the same reactions, E M noble e , E M e their equilibrium potentials and The equilibrium potentials can be estimated based on their standard values and the experimental conditions (temperature, metal ion concentration, ligand concentration and pH if relevant) while the exchange current densities and Tafel slopes can be determined experimentally.Equation (10) leads to the important conclusion that it is not only the thermodynamics but also the kinetics of a galvanic replacement process that is determined by the equilibrium potential difference of the two metals (in cases where both half-cell reactions are under pure kinetic control).(Note, however, that, since the concentrations of the metal ions involved change as the process proceeds (that of the more noble metal decreases while that of the other metal may increase), so do the equilibrium potentials (according to the Nernst equation) and as a result both the potential difference in Equation (10) and the galvanic replacement rate decrease in time.) In the case of relatively fast M dissolution rates and if M noble n+ deposition is under mass transfer control, then the galvanic current density/rate will be determined by M noble n+ mass transfer.For example, in the case of stationary solutions, the mass transfer controlled galvanic replacement current density can also be written (according to the Cottrell equation) as: while the decrease in M noble n+ concentration should be given by: where k m is the mass transfer coefficient, A is the substrate surface area and V the solution volume.It should be stressed that the above analysis strictly holds for large substrates of pure metal M of a known area A in contact with M noble n+ .It could therefore only serve as a qualitative predictor of galvanic replacement/deposition kinetics in cases that the process occurs through defects in oxide passivation layers or that the partial current density corresponding to M noble n+ reduction/deposition is sustained by the passivation partial current density of M dissolution through a homogeneous protective oxide via a high-field mechanism (see, for example, [15]).Nevertheless, in the latter case, if the rate is controlled by diffusion of M noble n+ to the entire passivated surface, then Equations ( 11) and ( 12) may still be applicable.

Metal Finishing, the Electronics Industry, and Metallurgical Applications
Galvanic replacement has been widely used in the printed circuit board (PCB) technology and the metal finishing industry in general [16][17][18][19], under the name of immersion plating.For example, tin/lead immersion coatings on Cu (from baths containing appropriate complexing agents) [20] and gold or silver immersion coatings on Cu (sometimes coupled with Ni electroless plating) [21][22][23] are common practice in PCB fabrication.Also, Zn and Cu immersion coatings on aluminum have been used for anticorrosion, primer plating, or decorative purposes [24][25][26].Finally, Cu immersion plating has been used for corrosion protection of stainless steel substrates [27].
In a less well-known application, galvanic deposition has been used in hydrometallurgy as a means of retrieving metals from their solutions onto appropriate metallic substrates [28] (in that case it is often classified as cementation).For example, Cu was being recovered from mining effluents using iron substrates as collectors [29].Similarly, it was recently shown that the decrease in biocide Ag ion concentration observed in potable water stored in stainless steel tanks used in space missions was due to Ag galvanic deposition on stainless steel, with simultaneous growth of oxides of the latter [30].

Ultrathin Film Applications
Galvanic replacement has also been applied to the formation of ultrathin (even atomically thin) epitaxial metal layers over a metal substrate, with the technique known as surface limited redox replacent (SLRR) (see the excellent reviews of [6,31]); such films have unique electronic, optical and catalytic properties that make them indispensable components of advanced electronic and optoelectronic devices.This process belongs to the larger family of surface-limited reactions (SLR) and more specifically to those that the first step involves the formation of a monolayer of a metal (most often) or hydrogen, sulphur or iodine atoms by means of their reductive or oxidative deposition at potentials more positive or negative respectively than that required for bulk deposition; these monolayers are known as uderpotential deposits (upd; see, for example, [32]).Historically, the first application of a SLR based on upd was for the preparation of epitaxial layers of semiconductor compounds (e.g., CdTe, GaAs) and was due to Stickney and co-workers [33][34][35][36], who established the technique as electrochemical atomic layer epitaxy (ECALE).However, the first application of SLRR to prepare a metal monolayer on a metal substrate by galvanic replacement of a pre-deposited metal upd was due to Brankovic et al. [37], who deposited Pt submonolayers, Pd monolayers and Ag bilayers on a Au (111) substrate by the galvanic replacement of a Cu upd monolayer (the difference in atomic layer coverage resulted from the use of Pt(IV), Pd(II), and Ag(I) solutions and the stoichiometry of the corresponding exchange reactions that involved Cu oxidation to Cu(II)).Subsequently, Weaver and co-workers [38] used successive applications of the two elementary steps (upd deposition and redox replacement) to create multilayer Pt coatings consisting of epitaxially deposited monolayers.Following these early works, Dimitrov and co-workers [39][40][41][42][43][44][45][46] and Stickney and co-workers [47][48][49][50][51][52][53] fully explored the technique for the preparation of a number of noble metal layers with atomic layer epitaxy.They further advanced practical aspects of the method that include automation of the process using a flow cell [51], a single-cell approach with careful control of upd deposition and ion concentrations [42], the use of adsorbed H as a sacrificial layer [44,53], the electroless deposition of the sacrificial upd layer by short circuiting the substrate with an appropriate counter electrode [45,46] and even some electrocatalytic applications [43,52].

Nanoparticle and Catalytic Layer Applications
Of the most recent applications of galvanic replacement in materials chemistry has been the preparation of bimetallic (often core-shell) or/and hollow nanoparticles, having useful optical, magnetic and/or catalytic properties, as recently reviewed in [54].One of the first relevant works involved the preparation of Au, Pt, and Pd nanoparticles by the replacement of Cu or Ag precursors encapsulated in organic dendrimers [55]; the thus-prepared Pt and Pd nanoparticles showed enhanced oxygen reduction electrochemical activity.Examples of other early papers that followed include the preparation of Pt-Co nanoparticles and the exploration of their magnetic properties [56], the preparation of Au-Ag, Pd-Ag, Au-Pd and Au-Cu bimetallic particles with modified plasmonic behavior [57] and the preparation of hollow Pt, Pd, and Au nanostructures using Ag nanoparticle or nanotube templates [58].Au nanocages prepared by galvanic replacement have also found applications in medicine for targeted cancer therapy and imaging, due to their localized surface plasmon resonance properties that allow heat conversion of Near-infrared (NIR) radiation (see, for example, [59,60]).
Following the elegant work of [55], there have been many reports on the preparation of bimetallic and/or hollow electrocatalysts and reviews of such catalysts used in fuel cells or/and other energy applications are given in [61,62] and [63] respectively.Among the most systematic series of works in the preparation, characterization and testing of electrocatalysts prepared by galvanic replacement/deposition one has to mention that of Adzic and co-workers (see, for example, [64][65][66]), Kokkinidis and Sotiropoulos and co-workers (see, for example, [67,68]), Podlovchenko and co-workers (see, for example, [69,70]) and Musiani and co-workers (see, for example, [71,72]).
In more detail, Kokkinidis and co-workers prepared Pt-modified Ti electrodes on freshly polished Ti substrates by galvanic deposition and used them as hydrogen evolution and oxygen reduction cathodes [73,74]; shortly afterwards they have prepared Pt-Cu and Pt-Pb modified electrodes for oxygen reduction, by partial galvanic replacement of Cu and Pd bulk metal adlayers [75,76].During the same time period, Adzic and co-workers prepared Pt-modified Ru electrodes on reduced Ru(0001) surfaces or Ru nanoparticles by galvanic deposition and used them as CO-tolerant anodes [77,78]; they were also the first to demonstrate the possibility of forming Pt, Pd, and Ag sub-monolayers or monolayers on Au(001) by the complete galvanic replacement of a Cu under-potentially deposited monolayer [37].Following these pioneering papers in the area, Adzic and co-workers embarked on a very systematic study of various poly-metallic systems of noble metal sub-monolayer/monolayer shells (mostly Pt and Pd) over a noble metal core (mostly Pd or Au), always using the Cu upd replacement method; in some cases they also produced mixed noble metal shells and mixed noble/less-noble metal cores (see, for example, [64][65][66][79][80][81][82][83][84][85][86][87][88][89][90][91][92][93][94][95][96][97]).Most of their work focused on the use of these catalysts as efficient oxygen reduction cathodes, although they also studied them as alcohol and CO anodes as well.They were the first to scale up this preparation route to practical carbon-supported catalysts [90], soon followed by other work [53,98,99].On the other hand, Sotiropoulos and co-workers continued the partial galvanic replacement of metal poly-layers (Cu, Pb, Fe, Co, Ni) by more noble metals (mainly Pt, Au, Ru, Ir) and used the resulting electrodes in a number of fuel cell and electrolysis relevant reactions (oxygen reduction, methanol oxidation, borohydride oxidation, hydrogen and oxygen evolution) [67,68,[100][101][102][103][104][105][106][107][108][109][110][111][112][113].In a similar research path, Podlovchenko and co-workers, prepared thin deposits of noble metals (mainly Pt, Pd, and Au) on a number of less noble metals (Cu, Mo, Bi) by their partial galvanic replacement while also adopting the Cu upd technique too [69,70,[114][115][116][117][118][119][120][121][122][123][124][125][126][127][128]; the resulting electrodes were tested mainly for methanol and formic acid oxidation.Musiani and co-workers prepared Ru, Ir, Rh, and Ag deposits on a number of supports made of Cu, Ni, or Fe alloys by the partial galvanic replacement of surface layers of the latter from the former [71,72,[129][130][131][132][133][134][135][136][137][138]; they used the modified electrodes under alkaline conditions mainly in cathodic reactions (hydrogen evolution, nitrate reduction, organics reduction) but also in some anodic reactions too (alcohol oxidations) and in gas-phase catalysis.Similar work on hydrogen evolution on Ru-or Ir-modified Ni has been reported by Trasatti and co-workers [139,140].Finally, there have been a few papers where galvanic replacement of metal ions from their oxides has been used to prepare mixed oxides or oxide overlayers over other oxides; such systems involved oxides and ions of elements with multiple oxidation states, such as Mn, Fe, Co, and Sn [141,142].

Characteristics of the Two Galvanic Replacement Approaches for Electrocatalyst Preparation
From the historical background given above it follows that the approaches for galvanic replacement are based in one of the following paths: either the complete replacement of a Cu, Pb, or H upd monolayer by the noble metal catalyst M noble (mainly due to the groups of Adzic, Brankovic, Dimitrov, and Stickney) or the partial replacement of surface layers of a metal M by layers of the more noble metal catalyst M noble (adopted in most other works).
A schematic representation of the first approach is depicted in Figure 1A, where, for the sake of simplicity, the noble metal species in solution is assumed to carry a net charge of +2 so that one of its atoms replaces one atom of Cu upd.
There are two main advantages of this method.First, it leads to minimal amounts of the desired noble metal catalyst as the latter is only present as a thin overlayer.Ideally this can be a monolayer or sub-monolayer, depending on reaction stoichiometry, if a direct exchange growth mechanism is operative, whereby each noble metal atom strongly interacts with upd atoms and deposits only on top of them [31].On the contrary, if a local cell growth mechanism prevails (i.e., noble metal atoms are deposited on each other while the upd metal dissolves from a nearby location) then a slightly thicker, incomplete layer is expected [31].In both cases, however, the relatively thin noble metal layer results in very high currents per mg of catalyst, thus decreasing significantly the cost.Second, the thin M noble overlayer is prone to significant modifications of its electronic properties by the underlying metals of the core, either via geometric or ligand effects [143,144].For sub-monolayers and nanoclusters, finite size effects have also to be taken into account due to the different strain exerted by the substrate to small particles from that to continuous films; in this way the electronic properties and catalytic activity modification can also be correlated to particle size [145][146][147].Adzic and co-workers have systematically explored these effects and reported some of the highest Pt utilization values for oxygen reduction and alcohol oxidation (see, for example, [64][65][66]94,95]) while also finely tuning the shell properties by using multi-metallic cores that contain apart from a second noble metal M ' noble other noble or less noble metals M (see, for example, [87,93,96]).Among potential drawbacks of the technique is the need for electrodeposition of the sacrificial Cu, Pb, or H layer on a second noble metal M '  noble support (upd formation requires potentiostatic control or an appropriate galvanic cell [45,46], as well as a noble metal substrate), which complicates the preparation setup and increases the cost of materials.(Note that the recently proposed chemical formation of adsorbed sacrificial H on Pd [53] eliminates the need for an electrochemical cell but restricts the use of the method to Pd substrates.)Also, the presence of a single monolayer of the most active catalyst components raises issues of stability (although no long-term performance deterioration has been reported [80,93]).
Catalysts 2017, 7, 80 7 of 32 underlying metals of the core, either via geometric or ligand effects [143,144].For sub-monolayers and nanoclusters, finite size effects have also to be taken into account due to the different strain exerted by the substrate to small particles from that to continuous films; in this way the electronic properties and catalytic activity modification can also be correlated to particle size [145][146][147].Adzic and co-workers have systematically explored these effects and reported some of the highest Pt utilization values for oxygen reduction and alcohol oxidation (see, for example, [64][65][66]94,95]) while also finely tuning the shell properties by using multi-metallic cores that contain apart from a second noble metal M ' noble other noble or less noble metals M (see, for example, [87,93,96]).Among potential drawbacks of the technique is the need for electrodeposition of the sacrificial Cu, Pb, or H layer on a second noble metal M ' noble support (upd formation requires potentiostatic control or an appropriate galvanic cell [45,46], as well as a noble metal substrate), which complicates the preparation setup and increases the cost of materials.(Note that the recently proposed chemical formation of adsorbed sacrificial H on Pd [53] eliminates the need for an electrochemical cell but restricts the use of the method to Pd substrates.)Also, the presence of a single monolayer of the most active catalyst components raises issues of stability (although no long-term performance deterioration has been reported [80,93]).A schematic representation of the second approach is depicted in Figure 1B, where, again, for the sake of simplicity, the noble metal (Mnoble) and less noble metal (M) replacement reaction is assumed to have a 1:1 stoichiometry ratio.In this case, however, a thicker shell of Mnoble is formed and the latter also enters the core of the catalyst, resulting in a Mnoble-M mixed core (evidence of that is presented in [102] and discussed in the next section).If the galvanic replacement step does not proceed to the completion of a full noble metal shell (as is the case of either slow or uneven processes) and there is no further catalyst treatment, then the surface of the material contains both Mnoble and M sites and can usually only be employed as a cathode under alkaline conditions (otherwise M will be leached); this is the case of catalytic layers produced on bulk Cu, Ni, and Fe alloy substrates and used as cathodes in aqueous media by Musiani and co-workers [129,132,134,137,138] and Trasatti and co-workers [139,140].However, if the starting material is made of nanoparticles of M then, choosing the appropriate preparation conditions (e.g., concentration of Mnoble ions, complexing agents, time of reaction) or/and catalyst post-treatment (e.g., acid pepsis) a variety of interesting nanoparticle structures can be obtained (including hollow Mnoble or Mnoble-M nanoparticles), as reviewed in [54].Finally, if the initially formed Mnoble-M layers or particles are prepared on an electrode substrate or formed into an electrode and are exposed to anodic potentials then, by selective anodic dissolution of uncovered areas, an Mnoble shell-(Mnoble-M) A schematic representation of the second approach is depicted in Figure 1B, where, again, for the sake of simplicity, the noble metal (M noble ) and less noble metal (M) replacement reaction is assumed to have a 1:1 stoichiometry ratio.In this case, however, a thicker shell of M noble is formed and the latter also enters the core of the catalyst, resulting in a M noble -M mixed core (evidence of that is presented in [102] and discussed in the next section).If the galvanic replacement step does not proceed to the completion of a full noble metal shell (as is the case of either slow or uneven processes) and there is no further catalyst treatment, then the surface of the material contains both M noble and M sites and can usually only be employed as a cathode under alkaline conditions (otherwise M will be leached); this is the case of catalytic layers produced on bulk Cu, Ni, and Fe alloy substrates and used as cathodes in aqueous media by Musiani and co-workers [129,132,134,137,138] and Trasatti and co-workers [139,140].However, if the starting material is made of nanoparticles of M then, choosing the appropriate preparation conditions (e.g., concentration of M noble ions, complexing agents, time of reaction) or/and catalyst post-treatment (e.g., acid pepsis) a variety of interesting nanoparticle structures can be obtained (including hollow M noble or M noble -M nanoparticles), as reviewed in [54].Finally, if the initially formed M noble -M layers or particles are prepared on an electrode substrate or formed into an electrode and are exposed to anodic potentials then, by selective anodic dissolution of uncovered areas, an M noble shell-(M noble -M) core structure may be formed (see lower part of Figure 1B) as presented in [100][101][102][103][104][105][106][107][108][109][110]; these catalytic electrodes can then work as anodes and/or in acid solutions due to the protective M noble shell and shall be denoted as M noble (M) hereafter.Note that, although this post-treatment is essentially the same as the electrochemical de-alloying introduced and established by Strasser and co-workers (see, for example, the recent review of [148]), the core of the particle in the case of galvanic replacement is formed via a room/moderate temperature process (and results in partial M noble -M alloying [100][101][102][103][104][105][106][107][108][109][110]), whereas that of the latter method is prepared by elevated temperature annealing or hydrothermal routes (and results in a well-defined core alloy).Advantages of the second approach for galvanic replacement (when coupled with the formation of a noble metal shell) over the one based on a upd layer include the expected higher stability of the thicker noble metal skin, the fact that electrodeposition of the upd monolayer on a second noble metal is not needed and the use of a less expensive and more reactive metal core (e.g., Cu, Fe, Co, Ni, Pb); that core, consisting of atoms much smaller and much less electronegative than the noble metal, is also expected to exert a stronger effect on the properties of the latter [143,144].Drawbacks come from the increased thickness of the noble metal skin, which leads to a cost increase and an attenuation of the core electronic and geometric effects on the outmost noble metal layers.

Types of Support, Methods of Less Noble Metal Preparation/Deposition and Catalyst Characteristics
Strictly speaking, all electrocatalysts have to be supported on an electronic conductor or semiconductor support that is either in the form of particles and/or a solid block that acts as a current collector.However, we shall distinguish here between electrocatalysts prepared by galvanic replacement in those that the support material is present during their preparation (supported catalysts) and those that are initially formed in the absence of a support material and subsequently mixed with or applied onto the support.
In basic research studies, where the focus has been the feasibility and the mechanism of the method as well as the morphology and the interactions between the multi-metallic system components, well-characterized smooth electrode substrates (such as Au and Ru monocrystals [37,77] Ti and Au discs [73][74][75][76] and glassy carbon, GC [67,[100][101][102][103][104][105][106]109,113]) have been used.Also, in those cases, the preferred preparation method of the sacrificial upd monolayer or of the precursor less noble metal M multi-layers has been that of electrodeposition (offering accurate control of the coverage and/or thickness and morphology of the deposits of M). Figure 2 presents a schematic diagram of the expected structure of bimetallic Pt(M) particles prepared by the partial galvanic replacement of surface layers of M from their thin bulk electrodeposits on a smooth GC electrode.
Two situations can be envisased: the first is when the process is fast and a protective thin Pt overlayer is quickly formed, leaving a continuous M underlayer; the second is when the process is slow and M has time to completely dissolve from certain sites, leaving Pt(M) particles on the GC surface, with a relatively thick Pt shell and Pt penetration into the particle core.One of the parameters that may affect the rate of the process and hence the morphology of the deposit is, according to the reasoning of Section 1 above, the E M noble e − E M e difference (in this example the E Pt e − E M e difference).The larger this difference, the more likely is the morphology of the deposit to be that depicted at the top of Figure 2; the smaller, the more likely it is to look like that at the bottom of the same figure .method as well as the morphology and the interactions between the multi-metallic system components, well-characterized smooth electrode substrates (such as Au and Ru monocrystals [37,77] Ti and Au discs [73][74][75][76] and glassy carbon, GC [67,[100][101][102][103][104][105][106]109,113]) have been used.Also, in those cases, the preferred preparation method of the sacrificial upd monolayer or of the precursor less noble metal M multi-layers has been that of electrodeposition (offering accurate control of the coverage and/or thickness and morphology of the deposits of M). Figure 2 presents a schematic diagram of the expected structure of bimetallic Pt(M) particles prepared by the partial galvanic replacement of surface layers of M from their thin bulk electrodeposits on a smooth GC electrode.Co being lower than E e Cu (see Table 1 above) and with the previous argument, the Pt(Cu) deposit is much rougher than the Pt(Co) one (the difference in the morphology of the final products is much more pronounced than that of the original Cu and Co deposits).Also the composition change of the as prepared Pt(Cu) catalyst when activated by anodic polarization, points to the anodic dissolution of many uncovered Cu areas, which in turn indicates the slow/incomplete reaction of Cu layers with Pt.However, following repetitive potential scans between the hydrogen and oxygen evolution potential limits, both Pt(M) (and Au(M)) catalysts obtained a protective Pt (or Au) shell by the dissolution of uncovered M areas and the rearrangement/surface segregation of Pt (or Au), as indicated by the fact that the surface electrochemistry of these layers in acid is very similar qualitatively to that of bulk Pt (or Au) (see Figure 4A,B [102]).
Catalysts 2017, 7, 80 9 of 32 Two situations can be envisased: the first is when the process is fast and a protective thin Pt overlayer is quickly formed, leaving a continuous M underlayer; the second is when the process is slow and M has time to completely dissolve from certain sites, leaving Pt(M) particles on the GC surface, with a relatively thick Pt shell and Pt penetration into the particle core.One of the parameters that may affect the rate of the process and hence the morphology of the deposit is, according to the reasoning of Section 1 above, the E E − difference).The larger this difference, the more likely is the morphology of the deposit to be that depicted at the top of Figure 2; the smaller, the more likely it is to look like that at the bottom of the same figure.
Figure 3 shows such Pt(Cu) and Pt(Co) bimetallic layers formed by the galvanic replacement of electrodeposited Cu or Co layers on GC by Pt [101,103].It can be seen that, in accordance with Ee Co being lower than Ee Cu (see Table 1 above) and with the previous argument, the Pt(Cu) deposit is much rougher than the Pt(Co) one (the difference in the morphology of the final products is much more pronounced than that of the original Cu and Co deposits).Also the composition change of the as prepared Pt(Cu) catalyst when activated by anodic polarization, points to the anodic dissolution of many uncovered Cu areas, which in turn indicates the slow/incomplete reaction of Cu layers with Pt.However, following repetitive potential scans between the hydrogen and oxygen evolution potential limits, both Pt(M) (and Au(M)) catalysts obtained a protective Pt (or Au) shell by the dissolution of uncovered M areas and the rearrangement/surface segregation of Pt (or Au), as indicated by the fact that the surface electrochemistry of these layers in acid is very similar qualitatively to that of bulk Pt (or Au) (see Figure 4A,B [102]).Following the macroscopic-electrochemical proof of a Pt or Au skin in these Pt(M) and Au(M) bimetallics, the question arising is whether these metals are restricted to the outer layers of the bimetallic layers or they enter the core of the catalyst too.The answer may be given by sputter-etch Auger Electron Spectroscopy (AES), which can provide the elemental composition depth profile of the deposits.Figures 5A, B show such profiles for Pt(Cu) and Pt(Ni) [67,102]; it can be seen that, although Pt is preferentially located on the outer surface, it also penetrates the core of the deposit, down to the substrate (in line with a morphology more reminiscent of that depicted at the bottom of Figure 2).However, the fact that a signal from M (Cu or Ni in the examples of Figure 5) could be obtained even before sputtering [67,102] means that the Pt outer shell is thin enough for the Auger electrons of M to escape and be detected; in fact, back of the envelope calculations [67], taking into account the inelastic mean free path of Auger electrons of the core metal (Cu, Ni, Co, Fe, Pb) and the atomic dimensions of Pt, point to the outer shell consisting of no more than four Pt monolayers.This is very important since any electronic and geometric effects of the underlayers of M on the Pt overlayers cannot be extended beyond a few monolayers [149].Finally, grazing incidence XRD experiments of these model deposits indicate a significant degree of Pt-M alloying within the bimetallic layers (deduced by a shift of the Pt peaks) [106].Following the macroscopic-electrochemical proof of a Pt or Au skin in these Pt(M) and Au(M) bimetallics, the question arising is whether these metals are restricted to the outer layers of the bimetallic layers or they enter the core of the catalyst too.The answer may be given by sputter-etch Auger Electron Spectroscopy (AES), which can provide the elemental composition depth profile of the deposits.Figure 5A,B show such profiles for Pt(Cu) and Pt(Ni) [67,102]; it can be seen that, although Pt is preferentially located on the outer surface, it also penetrates the core of the deposit, down to the substrate (in line with a morphology more reminiscent of that depicted at the bottom of Figure 2).However, the fact that a signal from M (Cu or Ni in the examples of Figure 5) could be obtained even before sputtering [67,102] means that the Pt outer shell is thin enough for the Auger electrons of M to escape and be detected; in fact, back of the envelope calculations [67], taking into account the inelastic mean free path of Auger electrons of the core metal (Cu, Ni, Co, Fe, Pb) and the atomic dimensions of Pt, point to the outer shell consisting of no more than four Pt monolayers.This is very important since any electronic and geometric effects of the underlayers of M on the Pt overlayers cannot be extended beyond a few monolayers [149].Finally, grazing incidence XRD experiments of these model deposits indicate a significant degree of Pt-M alloying within the bimetallic layers (deduced by a shift of the Pt peaks) [106].For practical applications, electrocatalysts have to be in nanoparticle form and, in many fuel cell applications, supported on high surface area carbons.Electrodeposition of the less noble metal M onto carbon or of the upd monolayer on noble-metal decorated carbon particles is one of the routes for preparing the precursor M/C or upd/M ' noble/C particles; these can be transformed to the bimetallic Mnoble(M)/C or Mnoble(M ' noble)/C by reaction with the galvanic replacement solution (see, for example, [90,119]).Simpler routes for sacrificial M metal deposition on C particles include chemical reduction of M ions by a reducing agent (e.g., NaBH4 or EG (ethylene glycol) in a slurry of the carbon powder (see, for example, [68,108,150]) and the more delicate electroless deposition of M on C [107,110].A schematic diagram of this preparation process of Pt(Cu)/C catalysts is depicted in Figure 6.For practical applications, electrocatalysts have to be in nanoparticle form and, in many fuel cell applications, supported on high surface area carbons.Electrodeposition of the less noble metal M onto carbon or of the upd monolayer on noble-metal decorated carbon particles is one of the routes for preparing the precursor M/C or upd/M ' noble /C particles; these can be transformed to the bimetallic M noble (M)/C or M noble (M ' noble )/C by reaction with the galvanic replacement solution (see, for example, [90,119]).Simpler routes for sacrificial M metal deposition on C particles include chemical reduction of M ions by a reducing agent (e.g., NaBH 4 or EG (ethylene glycol) in a slurry of the carbon powder (see, for example, [68,108,150]) and the more delicate electroless deposition of M on C [107,110].A schematic diagram of this preparation process of Pt(Cu)/C catalysts is depicted in Figure 6.For practical applications, electrocatalysts have to be in nanoparticle form and, in many fuel cell applications, supported on high surface area carbons.Electrodeposition of the less noble metal M onto carbon or of the upd monolayer on noble-metal decorated carbon particles is one of the routes for preparing the precursor M/C or upd/M ' noble/C particles; these can be transformed to the bimetallic Mnoble(M)/C or Mnoble(M ' noble)/C by reaction with the galvanic replacement solution (see, for example, [90,119]).Simpler routes for sacrificial M metal deposition on C particles include chemical reduction of M ions by a reducing agent (e.g., NaBH4 or EG (ethylene glycol) in a slurry of the carbon powder (see, for example, [68,108,150]) and the more delicate electroless deposition of M on C [107,110].A schematic diagram of this preparation process of Pt(Cu)/C catalysts is depicted in Figure 6.As pointed out by Podlovchenko and co-workers [119] and Sotiropoulos and co-workers [68,107,108], in such systems the noble metal (here Pt) is expected to deposit not only on sacrificial M islands (here Cu) but also on nearby bare C locations; this is because the substrate is an electronic conductor and electrons released by the dissolution of Cu can travel to nearby locations where Pt ions may be reduced and deposited as metallic Pt.Note that this effect is expected to occur at a lower extent in catalysts prepared via the Cu upd method since the whole process of replacing a single monolayer is restricted both in time and space; also, in principle, it should be absent if the support of M is an insulator or semiconductor (e.g., TiO 2 or WO 3 , as in [111,112,151]) or if the particles are prepared before the addition of the support (see, for example, [152]. Figure 7A,B presents Pt(Cu) and Pt(Ni) nanoparticles resulting from the (partial) galvanic replacement of electrolessly prepared Cu/C and Ni/C precursors [107,110].The TEM photographs reveal aggregates 10-20 nm large for Pt(Cu) and smaller than 10 nm for Pt(Ni), which are larger than those of the commercial Pt/C catalyst shown in Figure 7C.
Catalysts 2017, 7, 80 12 of 32 As pointed out by Podlovchenko and co-workers [119] and Sotiropoulos and co-workers [68,107,108], in such systems the noble metal (here Pt) is expected to deposit not only on sacrificial M islands (here Cu) but also on nearby bare C locations; this is because the substrate is an electronic conductor and electrons released by the dissolution of Cu can travel to nearby locations where Pt ions may be reduced and deposited as metallic Pt.Note that this effect is expected to occur at a lower extent in catalysts prepared via the Cu upd method since the whole process of replacing a single monolayer is restricted both in time and space; also, in principle, it should be absent if the support of M is an insulator or semiconductor (e.g., TiO2 or WO3, as in [111,112,151]) or if the particles are prepared before the addition of the support (see, for example, [152]. Figure 7A,B presents Pt(Cu) and Pt(Ni) nanoparticles resulting from the (partial) galvanic replacement of electrolessly prepared Cu/C and Ni/C precursors [107,110].The TEM photographs reveal aggregates 10-20 nm large for Pt(Cu) and smaller than 10 nm for Pt(Ni), which are larger than those of the commercial Pt/C catalyst shown in Figure 7C.XRD data for these [107,110] and similar [68,108] catalysts revealed an apparently Pt-rich alloy.The latter has usually a higher Pt content than that of the entire catalyst as estimated by EDS analysis [107,108,110], which could be the result of low degree of alloying (with pure Cu [107,108,110] pockets within the core of the catalyst).On the contrary, EDS and XRD analysis of the Pt(Cu) prepared in [68] points either to the coexistence of Pt pockets within the bimetallic particles or/and pure Pt deposits at nearby C locations (according to the reasoning of the previous paragraph).Incomplete alloying may also be the reason why no significant shift in Pt X-ray photoelectron XRD data for these [107,110] and similar [68,108] catalysts revealed an apparently Pt-rich alloy.The latter has usually a higher Pt content than that of the entire catalyst as estimated by EDS analysis [107,108,110], which could be the result of low degree of alloying (with pure Cu [107,108,110] pockets within the core of the catalyst).On the contrary, EDS and XRD analysis of the Pt(Cu) prepared in [68] points either to the coexistence of Pt pockets within the bimetallic particles or/and pure Pt deposits at nearby C locations (according to the reasoning of the previous paragraph).Incomplete alloying may also be the reason why no significant shift in Pt X-ray photoelectron spectroscopy (XPS) peaks has been observed [68,107,108,110].This, however, does not exclude the possibility of significant modification of the outer Pt skin electronic properties (at locations where it is thin enough) by the underlying core metal, since XPS may not be surface sensitive enough to probe the outer layers.Furthermore, as pointed out in [153], neither conventional XRD nor XPS can give decisive information about nanoparticles of a non-homogeneous composition.A review of the various characterization techniques for core-shell nanoparticles and the information they provide is due to Koper and co-workers [154].An excellent paper presenting the results of state-of-the-art microscopic and spectroscopic techniques (High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), STEM-electron energy loss spectroscopy (STEM-EELS), STEM-energy dispersive X-ray spectroscopy (STEM-EDS), Extended X-ray absorption fine structure (EXAFS)) for the characterization of bimetallic nanoparticles prepared by galvanic replacement is due to Adzic and co-workers [90] while the exciting real time observation of Pd deposition on Ag nanoparticles by in situ liquid-cell electron microscopy has been reported by Sutter and co-workers [155].
Finally, one should keep in mind that a large number of works are based on the separate preparation of the bimetallic particles without the presence of a support (which can be added at a second step).In that case, the precursor metal M or M-M ' noble nanoparticles are prepared by one of the common metal nanoparticle preparation techniques or their combination; these include hydrothermal reduction in the presence of polyols or NaBH 4 and high temperature hydrogen annealing (see, for example, [156,157]). Figure 7D shows such a Pt(Cu) nanoparticle, prepared by the galvanic replacement of Cu particles synthesized by hydrogen annealing of Cu oxides [152] and subsequently supported/mixed with C particles.

Oxygen Reduction
The oxygen reduction reaction (ORR) is of paramount importance for many electrochemical energy conversion applications such as those of fuel cells and batteries (see, for example, [158]).To improve the slow kinetics of oxygen reduction [159] in low and moderately elevated temperature fuel cells, Pt-based catalysts (the common choice in acid electrolytes) have been modified by the addition of other noble metals (e.g., Au, Pd etc.) or other transition metals (mainly Cr, Fe, Co, and Ni) to binary, ternary or even tertiary composite materials (see, for example, [160][161][162]).In line with this, core-shell polymetallic catalysts prepared by galvanic replacement have also been proposed.
Kokkinidis and co-workers and Sotiropoulos and co-workers have also tested their Pt(M) and Au(M) (M: Pb, Cu, Fe, Co, Ni) particulate deposits on GC as oxygen cathodes in acid and have found in some cases a moderate enhancement in their ORR activity per electroactive area [75,76,101,103,105]. (They have also pointed out the importance of the starting potential in voltammetric experiments in accessing ORR electrocatalytic activity at Pt-based electrodes [105].)Practical Pt(Cu)/C catalysts prepared by this approach on high surface area carbons also exhibited a moderate intrinsic activity enhancement and a comparable/slightly improved mass specific activity for ORR when compared to commercial Pt/C catalysts [68].The moderate improvement in ORR activity for this type of catalysts is due to the fact that the skin is expected to be thicker than that of the Adzic-type ones (see Figure 1); also, the core is expected to contain rather low quantities of M (see also the AES experimental evidence discussed in the previous section), so that the overall favorable modification of Pt properties is attenuated.
The modification of the catalytic properties of Pt overlayers (and other noble metals, M noble , that have a relatively large Wigner-Seitz atomic radius and high elegronegativity) when they are in contact with early transition metal M underlayers (of a smaller radius and electronegativity) has been widely attributed to the downward shift in Pt d-band energy centre, ε d .This shift has been predicted in the seminal work of Nørskov and co-workers [143,144] and is known to decrease the adsorption strength of the overlayers towards small species such as O, OH, H, and CO entities.It has been confirmed experimentally directly by high resolution ultraviolet photoelectron spectroscopy (UPS) measurements of d-electron energy [199,200] and indirectly by XPS measurements that confirmed a down-shift of core electron energy levels [201].
With respect to the higher ORR catalytic activity of Pt(M) electrodes, two interpretations have been offered.Ross, Markovic, Stamenkovic, and co-workers [199,200,[202][203][204][205][206] propose the reduction of Pt-O and Pt-OH (intermediates formed by oxygen adsorption) as the rate determining step of ORR.It is the weakening of the Pt-O bonds at Pt(M) systems that facilitates the removal of O adsorbed species and increases the rate of ORR.Contrary to these views, Watanabe and co-workers have found an increase in Pt-O coverage at Pt(Fe) layers [207], which they interpret by an increase in Pt d-electron back-donation from a higher Fermi level (with respect to the down-shifted e d and core energy levels [201]) to oxygen π* orbitals, a trend that increases the scission of O-O bonds and hence oxygen chemisorption and its reduction.

Methanol, Formic Acid, and Ethanol Oxidation
Direct methanol fuel cells, although not yet a viable solution for large scale applications, are of technological relevance for portable and micro-fuel cells in direct methanol fuel cells.Hence, a large number of papers are devoted to methanol electrooxidation in search for efficient anodes (see, for example, [208,209]).In search for efficient electrocatalysts that would remedy the Pt-based anode poisoning by a carbonaceous intermediate (most likely CO), many poly-metallic systems have been studied both in the form of ad-atom surface layers on Pt [210,211] and binary [212][213][214] or ternary [212,215] Pt alloys.It is now accepted that the best bi-metallic system is that of Pt-Ru but the high cost of Ru has made the identification of other metals M that increase Pt or Pt-Ru activity towards MOR very important.
Poly-metallic Pt-based catalysts prepared by galvanic replacement have been tested in methanol oxidation both for practical reasons (increased Pt utilization and modification of its electronic properties) and for fundamental research (the pure Pt shell of some of these systems ensures that only Pt-M electronic and lattice mismatch interactions will be operative and no synergistic or third-body effects of M will be present).
Pt monolayers/submonolayers prepared by the the Cu upd replacement technique onto Au, Ru, NbO 2 and PdCu [94,216,217] have been used for methanol oxidation by Adzic and co-workers and the resulting electrode surfaces have been proven more efficient than Pt.The same authors have also prepared electrodes by the replacement of Cu and Pb bulk deposits [218] by Pt or Pt-Ru layers and have prepared catalysts with an up to 10-fold increase in MOR activity with respect to commercial Pt-Ru/C catalysts.
Podlovchenko and co-workers have also tried the replacement of Cu upd by Pt on Pd and Au surfaces [125,126] as well as the partial replacement of Cu coatings (mainly electrodeposits) on carbon substrates [13,70,117,119]; depending on preparation conditions they found a moderate or small change in MOR activity at Pt(Cu) electrodes.However, the same authors have reported significant enhancement of MOR rates at Pt and Pd electrodes prepared by the partial replacement of H atoms from hydrogen-containing Mo bronzes [69,118,127,128,219].
Sotiropoulos and co-workers have found a very significant enhancement in the intrinsic MOR activity of Pt(Cu) layers on glassy carbon substrates with respect to Pt(Pb), Pt(Co), Pt(Ni), and pure Pt layers [67,100]).(The same workers have proven the applicability of the method to produce the MOR oxidation benchmark catalytic material of PtRu in the form of Pt-Ru (Ni) layers [109]).A similar enhancement of MOR activity (together with a noticeable shift of CO electrooxidation to less positive potentials) was recorded at Pt(Ni)/C and Pt(Cu)/C nanoparticles with respect to commercial Pt/C ones [107,108,152]; however, the mass specific activity of these catalysts is similar to that of commercial catalysts due to relatively large particles and rather small surface area.They have also found an increase of intrinsic MOR rates at Pt/TiO 2 and Pt/WO 3 layers or particles and a further increase upon UV and/or visible light illumination [111,151].
The increase in MOR catalytic activity that has been observed in most of the above multicomponent catalysts can be interpreted according to the effect(s) of the second metal/metal oxide on the catalyst properties, as reviewed by Ishikawa and co-workers [233].The second metal can increase the reaction rate by synergy (either disrupting the Pt reaction sites and the adsorption of poisonous intermediates-the "third-body effect"-or by hosting oxygenated species that remove the poisons) or via an electronic effect on Pt or Pd (either a "ligand" effect due to a difference in electronegativity or by a strain effect due to differences in atom size [143,144]).In those systems that the second metal/metal oxide is also on the surface (e.g., Pt-Ru, Pt-MO x , Pd-MO x , Pt-Au, Pt-TiO 2 , Pt-WO 3 catalysts) both the synergistic and electronic effects are operative (with the former expected to dominate) while in the case of a Pt (or Pd) skin, only the electronic effect is present.In the latter case, when the second metal is less electronegative and of a smaller atom radius than Pt (or Pd) then the down-shift of the e d level of Pt (or Pd) is expected to decrease their adsorption affinity for both the chemisorbed reactant (methanol) and the poisonous intermediate (CO); this in turn leads to an optimum adsorption affinity needed for maximum MOR rates, which has been claimed to be supplied by a Cu-rich core [67].

Hydrogen Evolution and Oxidation
The Hydrogen Evolution Reaction (HER) and the reverse reaction of Hydrogen Oxidation Reaction (HOR) are very important in electrochemical energy storage and conversion since H 2 produced by electrolysis of water in electrolyzers is the fuel feed in fuel cells [275,276].Both electrodes used for HER in alkaline media (e.g., Ni) and in acid media (e.g., Pt) have been modified by other metals via a galvanic replacement process that introduces a second metal to the system.
Trasatti and co-workers [139,140] and Musiani and co-workers [71,129,132] have modified Ni substrates with Ru and Ir nuclei and porous or three-dimensional Ni cathodes with Ir or Pt nuclei respectively.They have pointed out the critical effect of the state of Ru and Ir complexes, as well as the presence/absence of oxygen and pH, on the galvanic replacement process and have found that the modified cathodes showed enhanced HER performance in alkaline solutions (with respect to plain Ni).This finding is to be expected since these noble metals (Pt, Ir, Ru) are better catalysts for HER than Ni.Nevertheless, it is of practical importance that efficient HER cathodes with very low precious metal loadings can be prepared by a relatively simple galvanic replacement process.
Conflicting results have been reported for HER in acidic media at Pt(M) electrodes with a continuous Pt skin.In more detail, increased activity of Pt(Ag) hollow nanocubes [277] and Pt(Cu), Pd(Cu), or PtPd(Cu) deposits on Cu particles or foams [278][279][280][281] has been reported, but that was only based on current densities per substrate geometric area or per catalyst mass, hence they might not have been representative of the inherent activity of the systems.However, in a recent paper a trimetallic Pt(CuNi) catalyst anchored on carbon nanofibers was shown to possess both high mass specific activity and enhanced intrinsic activity for HER (per electroactive surface area) with respect to pure Pt [282].(One should note, however, that unusual Pt surface electrochemistry was recorded in that work and, also, Cu and Ni were found to be on the surface, possibly offering direct synergism to Pt rather than simple electronic modification.) On the other hand, Papadimitriou et al. found that Pt(M) catalytic layers (M: Fe, Co, Ni, Cu) [102] with a continuous Pt shell are characterized by decreased HER intrinsic activity.This is in line with results obtained at Pt(Ag) [283], Pt-Ni [284] and Pd(Fe) [285] electrodes.For Pt it can be explained in a qualitative manner by the expected down-shift of the Pt d-electron band energy level, e d (due to the presence of an underlayer containing smaller and less electronegative atoms), which in turn decreases Pt adsorption affinity [143,144]; the latter leads to a Pt(M)-H ads bond weakening as compared to Pt-H ads .Since HER activity is known to vary with metal-hydrogen bond strength in a volcano plot manner [286] and Pt lies very close to the top of the plot, any significant modification of its properties is likely to have an adverse effect on HER.The decreased hydrogen adsorption strength of Pt-Cu, Pt-Ni, Pt-Fe, and Pt-Co surface or near surface alloys has also been theoretically predicted based on DFT calculations [287][288][289][290]. Nevertheless, in [287] improved HER activity has been found for the Pt-Bi surface alloy, both theoretically and experimentally.The decreased HER activity of Pt(Cu), Pt(Fe), Pt(Co), and Pt(Ni) materials, although detrimental for their potential use as hydrogen producing cathodes, makes them candidate materials for hydrogenation reactions and for borohydride oxidation (see also below).
Finally, as far as HOR is concerned, Adzic and co-workers report increased mass and intrinsic activity at Pt submonolayers/shells deposited on Au or Ru substrates or nanoparticles [79,291] via the galvanic replacement of Cu upd layers.

Oxygen Evolution
Oxygen evolution reaction (OER) is important in many industrial applications as the anode reaction of the electrochemical cell and in particular in water electrolysis for hydrogen production at the cathode [292]; in cases that electrolysis is performed under acidic conditions (as in Solid Polymer Electrolyte, SPE, electrolyzers) metals/oxides more stable than Pt (such as Ir/IrO 2 ) are used.
There are very few papers testing catalysts prepared by galvanic replacement as OER anodes and these include Pt(Co) [293] and Ir(Cu) [294][295][296] bimetallic electrode systems that they show improved mass specific activity with respect to the precious metal (Pt or Ir).
Although catalyst particles with an IrO 2 shell and an Ir-Ni core have been reported by Strasser and co-workers to be excellent OER anodes, the initial Ir-Ni binary precursor was prepared on C or antimony-doped tin oxide by chemical reduction [297,298] or co-sputtering on Ti and Si substrates [299] (not by galvanic replacement); the precursor was subsequently electrochemically or thermally treated so that de-alloying and/or Ir surface segregation occurred, leading to the formation of an IrO 2 skin (due to anodization or heating).Papaderakis et al. [113] introduced IrO 2 (Ir-Ni) OER catalysts where the initial Ir(Ir-Ni) material was prepared by galvanic replacement and was further electro-oxidized to form an IrO 2 skin over the Ir(Ir-Ni) core (see Figure 8).Although catalyst particles with an IrO2 shell and an Ir-Ni core have been reported by Strasser and co-workers to be excellent OER anodes, the initial Ir-Ni binary precursor was prepared on C or antimony-doped tin oxide by chemical reduction [297,298] or co-sputtering on Ti and Si substrates [299] (not by galvanic replacement); the precursor was subsequently electrochemically or thermally treated so that de-alloying and/or Ir surface segregation occurred, leading to the formation of an IrO2 skin (due to anodization or heating).Papaderakis et al. [113] introduced IrO2 (Ir-Ni) OER catalysts where the initial Ir(Ir-Ni) material was prepared by galvanic replacement and was further electro-oxidized to form an IrO2 skin over the Ir(Ir-Ni) core (see Figure 8).[113].Copyright 2016, American Chemical Sociely).
They recorded moderately higher OER currents (per IrO2 electroactive area) than plain IrO2/Ir electrodes and attributed that to an electronic and geometric effect of core Ni to the IrO2 shell, resulting in a destabilization of Ir-OH bonds, which is the rds of OER.

Borohydride Oxidation
Although not likely to replace H2/O2 fuel cells in large-scale and automobile applications, direct borohydride fuel cells may find applications in micro-fuel cells [300,301].Pt has considerable electrocatalytic activity for borohydride oxidation (BOR), but it also causes chemical decomposition of NaBH4 at open circuit while its very low hydrogen evolution overpotential does not allow operation close to the borohydride standard potential (which is lower than that of H2); it also shows a low coulombic efficiency since the reaction is completed with the loss of four electrons per molecule.On the other hand, Au is an inferior electrocatalyst but its coulombic efficiency approaches the maximum value, corresponding to the eight-electron reaction pathway [104,106].Bimetallic Pt-Au systems are expected to combine the high kinetics of Pt with the high coulombic efficiency of Au [302][303][304].
The above discussion shows that there is room for the improvement of Pt (via suppression of borohydride direct hydrolysis and/or of hydrogen evolution and/or an increase of its coulombic efficiency) and of Au (via improvement of its kinetics).To that direction, bi-and tri-metallic systems have been tested and galvanic replacement has been one of the preparation methods proposed.In more detail, improved BOR electrocatalysts include Au-Ni surface layers that have been prepared by selective replacement of Zn from Zn-Ni precursor layers [305], Pt(Co) deposits formed on Co3O4 nanosheets [306], and hollow Pt nanospheres via the replacement of sacrificial Co nanoparticles [307].Tegou et al. [106] have reported that platinized nickel or cobalt layers, Pt(Ni) and Pt(Co), show (Reproduced with permission from [113].Copyright 2016, American Chemical Sociely).
They recorded moderately higher OER currents (per IrO 2 electroactive area) than plain IrO 2 /Ir electrodes and attributed that to an electronic and geometric effect of core Ni to the IrO 2 shell, resulting in a destabilization of Ir-OH bonds, which is the rds of OER.

Borohydride Oxidation
Although not likely to replace H 2 /O 2 fuel cells in large-scale and automobile applications, direct borohydride fuel cells may find applications in micro-fuel cells [300,301].Pt has considerable electrocatalytic activity for borohydride oxidation (BOR), but it also causes chemical decomposition of NaBH 4 at open circuit while its very low hydrogen evolution overpotential does not allow operation close to the borohydride standard potential (which is lower than that of H 2 ); it also shows a low coulombic efficiency since the reaction is completed with the loss of four electrons per molecule.On the other hand, Au is an inferior electrocatalyst but its coulombic efficiency approaches the maximum value, corresponding to the eight-electron reaction pathway [104,106].Bimetallic Pt-Au systems are expected to combine the high kinetics of Pt with the high coulombic efficiency of Au [302][303][304].
The above discussion shows that there is room for the improvement of Pt (via suppression of borohydride direct hydrolysis and/or of hydrogen evolution and/or an increase of its coulombic efficiency) and of Au (via improvement of its kinetics).To that direction, bi-and tri-metallic systems have been tested and galvanic replacement has been one of the preparation methods proposed.In more detail, improved BOR electrocatalysts include Au-Ni surface layers that have been prepared by selective replacement of Zn from Zn-Ni precursor layers [305], Pt(Co) deposits formed on Co 3 O 4 nanosheets [306], and hollow Pt nanospheres via the replacement of sacrificial Co nanoparticles [307].Tegou et al. [106] have reported that platinized nickel or cobalt layers, Pt(Ni) and Pt(Co), show a lower BOR onset potential than pure Pt (see Figure 9A) and attributed this to the combined effect of increased chemical hydrolysis of borohydride to hydrogen that is further electrooxidized and hindered electrochemical hydrogen evolution (which increases the rest potential with respect to the thermodynamic equilibrium potential of the reaction).The latter part of the interpretation is in complete accordance with the experimental finding of HER suppression at the same type of Pt(M) electrodes discussed in Section 5.3.The same authors have managed to prepare a tri-metallic PtAu(Ni) electrode of a mixed Pt-Au surface [104], exhibiting the corresponding combination of good catalytic activity and coulombic efficiency (see Figure 9B).XRD measurements confirmed complete PtAu alloying, while surface and BOR electrochemistry proved the desired co-existence of Pt and Au on the surface (i.e., the absence of segregation); note that the preparation of PtAu surface alloys is a very difficult task since the two metals are not expected to form a single phase at non-elevated temperatures [308] and, on the other hand, Au will segregate on the surface at elevated temperatures [309].

Other Cathodic Reactions
The use of electrocatalysts of the Mnoble (M) type, prepared by galvanic replacement, as cathodes in neutral or alkaline media, offers the possibility of using electrodes with nuclei of Mnoble on an M substrate without the need for complete coverage-protection of the latter, since M (e.g., Ni, Cu, stainless steel) will be stable under these conditions.This means that there is no need for electrochemically dissolving any uncovered areas when M is in the form of layers or that bulk M substrates (plates, three-dimensional electrodes, etc.) can be used.
Most of the work in that direction is due to Musiani and co-workers.Apart from using Ir(Ni) and Pt(Ni) cathodes for HER [71,129,132], they also expanded their studies to environmental remediation and electroanalysis.In more detail, they proposed the use of Rh(Ni) catalytic layers on Ni foams as a cathode for efficient nitrate removal by its reduction [134] and the use of Rh(Cu) layers on Cu porous electrodeposits as sensing cathodes for nitrate and nitrite detection [138].They have also introduced Ag(Ni) cathodes for the reductive dechlorination of pesticide pollutants [137].
A similar cathodic reaction of an electrosynthetic interest that has been carried out at Ag(Ni) electrodes is that of the cathodic cleavage of the C-Br bond.Vanrenterghem et al. [310] have proven that Ag modified Ni layers and Ni foams show both higher intrinsic catalytic activity towards benzyl-bromide reduction and higher yields of toluene production during its bulk electrolysis.The same authors have managed to prepare a tri-metallic PtAu(Ni) electrode of a mixed Pt-Au surface [104], exhibiting the corresponding combination of good catalytic activity and coulombic efficiency (see Figure 9B).XRD measurements confirmed complete PtAu alloying, while surface and BOR electrochemistry proved the desired co-existence of Pt and Au on the surface (i.e., the absence of segregation); note that the preparation of PtAu surface alloys is a very difficult task since the two metals are not expected to form a single phase at non-elevated temperatures [308] and, on the other hand, Au will segregate on the surface at elevated temperatures [309].

Other Cathodic Reactions
The use of electrocatalysts of the M noble (M) type, prepared by galvanic replacement, as cathodes in neutral or alkaline media, offers the possibility of using electrodes with nuclei of M noble on an M substrate without the need for complete coverage-protection of the latter, since M (e.g., Ni, Cu, stainless steel) will be stable under these conditions.This means that there is no need for electrochemically dissolving any uncovered areas when M is in the form of layers or that bulk M substrates (plates, three-dimensional electrodes, etc.) can be used.
Most of the work in that direction is due to Musiani and co-workers.Apart from using Ir(Ni) and Pt(Ni) cathodes for HER [71,129,132], they also expanded their studies to environmental remediation and electroanalysis.In more detail, they proposed the use of Rh(Ni) catalytic layers on Ni foams as a cathode for efficient nitrate removal by its reduction [134] and the use of Rh(Cu) layers on Cu porous electrodeposits as sensing cathodes for nitrate and nitrite detection [138].They have also introduced Ag(Ni) cathodes for the reductive dechlorination of pesticide pollutants [137].
A similar cathodic reaction of an electrosynthetic interest that has been carried out at Ag(Ni) electrodes is that of the cathodic cleavage of the C-Br bond.Vanrenterghem et al. [310] have proven that Ag modified Ni layers and Ni foams show both higher intrinsic catalytic activity towards benzyl-bromide reduction and higher yields of toluene production during its bulk electrolysis.

Conclusions/Future Avenues
During the last two decades galvanic replacement has been established as a viable method for the preparation of noble metal (M noble )-based electrocatalysts.Advantages, when compared to the most common high temperature annealing or hydrothermal techniques, include: (i) It is a moderate temperature method, resulting in energy savings and most importantly in limited particle aggregation (thus increased surface area).(ii) It can lead (especially via its Cu, Pb, or H upd replacement variant) to a minimization of noble metal loadings in the polymetallic catalyst, since the noble metal can preferably (or even exclusively) be located at the outer layer(s).(iii) It is a relatively simple method that involves few steps and chemicals (especially in its variant where a bulk metallic substrate is modified by the more noble metal by simply immersing the former into an ionic solution of the latter).
A critical evaluation of the achievements and the potential of the three main variants of the method namely, the replacement of upd monolayers by M noble , the direct replacement of surface layers of nanoparticles of a metal M and the decoration of bulk structures of the latter with nuclei of M noble , can be summarized as follows: (i) The galvanic replacement of upd monolayers is perhaps the most mature method for preparing commercial catalysts and has shown remarkable noble metal mass utilization, especially for oxygen reduction and the Pt(Pd) system.The only foreseen hurdle is moving the catalyst preparation industry to use the apparatus needed for the electrodeposition of the sacrificial metal upd onto a fluidized electrode of M or M/C particles, that will be subsequently transformed to M noble (M) or M noble (M)/C particles.(The alternative, of treating the substrate with hydrogen to create the sacrificial H layers, can only find application for a few metal substrates, e.g., Pd.) (ii) The direct deposition of M noble on M or M/C nanoparticles, despite giving promising results especially for methanol and CO oxidation at Pt(Cu) electrodes, still suffers from rather low noble metal mass utilization.This is due to the penetration of M noble into the M core, the need for use of rather large M precursor particles (thus increasing surface area) and, for M/C particles, the simultaneous deposition of M noble on C. Ways to move forward with this variant will include the preparation of initially unsupported M noble (M) nanoparticles with a thin outer shell and, possibly, a hollow interior too.This will require standardizing existing nanomaterials preparation routes and testing their products as electrocatalysts.(iii) The direct deposition of M noble on M bulk substrates is a straightforward approach that can find applications in alkaline and neutral media such as hydrogen production via water electrolysis and organic electrosynthesis.Future avenues should include the use of inexpensive substrates (e.g., stainless steel) and the screening of many organic reactions.
In conclusion, the present review has shown that galvanic replacement has been established as a simple and cost-effective method for the production of catalysts for a number of important electrochemical applications.

Figure 1 .
Figure 1.Schematic representation of the mechanism of the two main approaches for galvanic replacement at an atomic level: (A) Replacement of Cu upd monolayer, formed on a noble metal M ' noble substrate (that may also contain another metal, Mother), by the noble metal of interest, Mnoble; (B) replacement of atoms of a reactive metal substrate, M, by the noble metal of interest, Mnoble, followed by electrochemical etching/anodization.

Figure 1 .
Figure 1.Schematic representation of the mechanism of the two main approaches for galvanic replacement at an atomic level: (A) Replacement of Cu upd monolayer, formed on a noble metal M 'noble substrate (that may also contain another metal, M other ), by the noble metal of interest, M noble ; (B) replacement of atoms of a reactive metal substrate, M, by the noble metal of interest, M noble , followed by electrochemical etching/anodization.

Figure 2 .
Figure 2. Schematic representation of the morphology of "platinized" M deposits, Pt(M), prepared by the galvanic replacement of continuous M layers deposited on smooth glassy carbon, GC, substrate: Continuous Pt(M) layer for fast replacement rates (top part); Particulate Pt(M) deposit for slow replacement rates (bottom part).

Figure 2 .
Figure 2. Schematic representation of the morphology of "platinized" M deposits, Pt(M), prepared by the galvanic replacement of continuous M layers deposited on smooth glassy carbon, GC, substrate: Continuous Pt(M) layer for fast replacement rates (top part); Particulate Pt(M) deposit for slow replacement rates (bottom part).

Figure 3
Figure 3 shows such Pt(Cu) and Pt(Co) bimetallic layers formed by the galvanic replacement of electrodeposited Cu or Co layers on GC by Pt [101,103].It can be seen that, in accordance with E eCo being lower than E e Cu (see Table1above) and with the previous argument, the Pt(Cu) deposit

Figure 3 .
Figure 3. Scanning Electron Microscope (SEM) micrographs of Cu and Co electrodeposits on a GC substrate, Pt(Cu), and Pt(Co) layers formed via galvanic replacement and the same layers after exposure to positive potentials for electrochemical etching of uncovered Cu and Co (Reproduced with permission from [101].Copyright 2007, Elsevier; Reproduced with permission from [103].Copyright 2008, Elsevier).First row, left to right: Cu, Pt(Cu), activated Pt(Cu); Bottom row, left to right: Co, Pt(Co), activated Pt(Co).The composition depicted corresponds to the atomic percentage of Cu or Co in the bimetallic Pt(Cu) or Pt(Co) layers, as measured by Energy Dispersive X-ray Spectroscopy (EDS).

Figure 3 .
Figure 3. Scanning Electron Microscope (SEM) micrographs of Cu and Co electrodeposits on a GC substrate, Pt(Cu), and Pt(Co) layers formed via galvanic replacement and the same layers after exposure to positive potentials for electrochemical etching of uncovered Cu and Co (Reproduced with permission from [101].Copyright 2007, Elsevier; Reproduced with permission from [103].Copyright 2008, Elsevier).First row, left to right: Cu, Pt(Cu), activated Pt(Cu); Bottom row, left to right: Co, Pt(Co), activated Pt(Co).The composition depicted corresponds to the atomic percentage of Cu or Co in the bimetallic Pt(Cu) or Pt(Co) layers, as measured by Energy Dispersive X-ray Spectroscopy (EDS).

Figure 5 .
Figure 5. Sputter-etch Auger Electron Spectroscopy (AES) analysis of Pt-based bimetallic layers prepared by galvanic replacement: (A) Depth profile of a Pt(Cu) deposit (Reproduced with permission from [102].Copyright 2007, Elsevier); (B) depth profile of a Pt(Ni) deposit (Adapted to exclude results of the first layer due to adventitious C contamination, with permission from [67].Copyright 2010, American Chemical Society).

Figure 6 .
Figure 6.Schematic representation of the formation and morphology of "platinized" Cu nanoparticles supported on carbon particles, Pt(Cu)/C, prepared by galvanic replacement.

Figure 5 .
Figure 5. Sputter-etch Auger Electron Spectroscopy (AES) analysis of Pt-based bimetallic layers prepared by galvanic replacement: (A) Depth profile of a Pt(Cu) deposit (Reproduced with permission from [102].Copyright 2007, Elsevier); (B) depth profile of a Pt(Ni) deposit (Adapted to exclude results of the first layer due to adventitious C contamination, with permission from [67].Copyright 2010, American Chemical Society).

Catalysts 2017, 7 , 80 11 of 32 Figure 5 .
Figure 5. Sputter-etch Auger Electron Spectroscopy (AES) analysis of Pt-based bimetallic layers prepared by galvanic replacement: (A) Depth profile of a Pt(Cu) deposit (Reproduced with permission from [102].Copyright 2007, Elsevier); (B) depth profile of a Pt(Ni) deposit (Adapted to exclude results of the first layer due to adventitious C contamination, with permission from [67].Copyright 2010, American Chemical Society).

Figure 6 .
Figure 6.Schematic representation of the formation and morphology of "platinized" Cu nanoparticles supported on carbon particles, Pt(Cu)/C, prepared by galvanic replacement.

Figure 6 .
Figure 6.Schematic representation of the formation and morphology of "platinized" Cu nanoparticles supported on carbon particles, Pt(Cu)/C, prepared by galvanic replacement.

Figure 8 .
Figure 8. Schematic representation of the formation and morphology of IrO2-coated Ir(Ni)/GC deposits, prepared by galvanic replacement of Ni by Ir, followed by anodic growth of IrO2 on the Ir shell.(Reproduced with permission from[113].Copyright 2016, American Chemical Sociely).

Figure 8 .
Figure 8. Schematic representation of the formation and morphology of IrO 2 -coated Ir(Ni)/GC deposits, prepared by galvanic replacement of Ni by Ir, followed by anodic growth of IrO 2 on the Ir shell.(Reproducedwith permission from[113].Copyright 2016, American Chemical Sociely).