Composite Hypo-Hyper-d-Intermetallic and Interionic Phases as

3030

J. Phys. Chem. B 2006, 110, 3030-3042

Composite Hypo-Hyper-d-Intermetallic and Interionic Phases as Supported Interactive Electrocatalysts S. G. Neophytides,† K. Murase,‡ S. Zafeiratos,† G. Papakonstantinou,† F. E. Paloukis,† N. V. Krstajic,§ and M. M. Jaksic*,†,§ Institute of Chemical Engineering and High Temperature Chemical Processes, FORTH, and Department of Chemistry, UniVersity of Patras, Patras, Greece, Department of Materials Science and Engineering, Kyoto UniVersity, Kyoto, Japan, and Faculty of Technology and Metallurgy, UniVersity of Belgrade, Belgrade, Serbia & Montenegro ReceiVed: July 27, 2005; In Final Form: December 7, 2005

Interactive, strong interbonding and highly electron conductive nonstoichiometric titanium suboxide catalytic supports, Magneli phases (TinO2n-1, on average Ti4O7), have been used in the electrocatalysis of hydrogen (HELR) and oxygen (OELR) electrode reactions with remarkable consequences and advanced achievements. The theory of hypo-hyper-d-interelectronic bonding of transition metal ions and atoms has been employed for selective ordered grafting and shown to stay in the core of the strong metal-support interaction (SMSI) in heterogeneous catalysis and electrocatalysis, and thereby the substantial cause for the improved synergistic activity of composite (electro)catalysts. The same fundament has been the thermodynamic basis for the thermal production of symmetric intermetallic Laves type phases of nanostructured electrocatalysts, in particular the ones with higher oxophilic properties of hypo-d-elements. Remarkably advanced in electrocatalytic activity, highly monatomically dispersed deposits of Pt upon Magneli phases are shown to be unique and highly promising electrocatalysts for the cathodic oxygen reduction (ORR). Nanostructured Au upon a thin nanocrystalline film of anatase titania has been confirmed by X-ray photoelectron spectroscopy (XPS) as a typical classical paradigm of the SMSI, and at the same time affording the basis for gold with strained d-orbitals, as the reversible hydrogen electrode. Magneli phases have been shown to be the best electrocatalytic supports with unique properties both for low temperature PEM fuel cells (LT PEM FCs) with pronounced CO tolerance and water electrolysis in membrane type hydrogen generators.

1.0. Introduction The difference between catalysis and electrocatalysis is that each individual nanoparticle in common heterogeneous catalysis takes the role of an independent microscale reactor and additively contributes to the overall integral catalytic effect, while in electrocatalysis all such individual cluster centers have to be properly connected and integrated by the current collector to be included in the whole electrocatalytic process. However, experimental evidence shows that a remarkable percentage, sometimes more than 30%, fails to be properly connected when on carbon supports (Vulcan XC-72). Thus, the main purpose of the present paper is to introduce a specific, novel type of supported individual or composite nanostructured hypo-hyperd-intermetallic phases of transition elements and interionic electrocatalysts deposited on interactive and conductive oxide carriers. Such Magneli phase materials show improvements in their synergistic catalytic activity, CO tolerance, electronic conductivity, and current collection. An additional aspect would be to show the interactive hypo-hyper-d-interelectronic bonding nature of the so-called strong metal-support interaction (SMSI) in heterogeneous catalysis and electrocatalysis, resulting in an * Corresponding author. E-mail: [email protected]. Present address: Institute of Chemical Engineering and High-Temperature Chemical Processes, FORTH, P.O. Box 1414, Stadiou Road, Platani, GR-265 00 Patras, Greece. Phone: +30-2610-965272. Fax: +30-2610-965223. † University of Patras. ‡ Kyoto University. § University of Belgrade.

advanced overall effect on the entire catalytic process. In fact, the overall target would be to combine and advance the synergetic electrocatalytic and oxophilic contribution to CO tolerance in the hydrogen electrode reactions (HELRs), both by the intermetallic and SMSI on the basis of the hypo-hyperd-interelectronic, basically interionic, interactive bonding effects. 1.1. Hypo-Hyper-d-Interatomic Bonding Theory and Corresponding Relations. The Brewer intermetallic bonding theory,1-4 along with Friedel d-d-intermetallic correlations5,6 and the Miedema7 theory of intermetallic bonding, thermodynamically predicts that whenever metals of the left half of the transition series or hypo-d-electronic elements, having empty or half-filled vacant d-orbitals (bonding d-band, lower d-valence, d1-d5), are alloyed with metals of the right half of the transition series or hyper-d-electronic elements (or antibonding d-band, d6-d19), having internally paired d-electrons not available for bonding in the pure metal, intermetallic phases and stoichiometric compounds usually of extraordinary stability occur, with the latter increasing from the 3d level to the 5d level8-10 and in general with the exposure of d-orbitals. The correlation between interactive cohesive bonding (Gibbs free enthalpy of formation) and their electrocatalytic activity, for both cathodic hydrogen evolution (HER) and anodic oxidation reactions (HOR), has long ago been pointed out.8,9 Such a theoretical statement and experimental evidence primarily concern the pronounced catalytic synergism characteristic of the symmetric Laves phases, both in their optimal electronic constitution and in their corresponding interrelating crystal structure.11

10.1021/jp0541415 CCC: $33.50 © 2006 American Chemical Society Published on Web 01/27/2006

Supported Interactive Electrocatalysts

J. Phys. Chem. B, Vol. 110, No. 7, 2006 3031

Figure 1. Electrocatalytic activities of various intermetallic phases along the Zr-Ni phase diagram (curve 1) for the HER, taken as the exchanged current (jo, closed circles, curve 3) and relative current density changes (j, closed circuits, curve 2) at constant overvoltage (-0.2 V).

For individual transition metals, the logarithm of exchange current density (log jo),12 or polarization (overpotential) at constant current density, and vice versa, the current density at constant overpotential,13 is a measure of electrocatalytic activity for the HER. The work function (Φ)14 and electronegativity (φ*) and/or chemical potential,15,16 as the main electrocatalytic parameters, follow the same asymmetric volcano plot shapes, with maxima at the same positions (Ni, Pd, Pt) along the periodic table.8-11 With respect to periodicity, the higher the exposure of the d-orbital in space, the higher electrocatalytic activity for the HELR. Such a state of experimental evidence along with the present theoretical model enables the so-called “one-to-one” correlation among them, thereby also leading to the fundamental linear Trasatti relation of far-reaching consequences in electrode kinetics and electrocatalysis,17,18

log jo ∝ R∆G(M-H)/kBT ∝ Rr∆Φ/kBT

(1)

where ∆G(M-H) ) µH is the gain in the free energy associated with adsorption of one hydrogen atom, r is the slope of the experimental linear dependence (log jo vs ∆Φ), kB is the Boltzmann constant, and R = 0.5, the symmetry factor in kinetic relations. This particular relation brings into the linear interdependence of the logarithm of the exchange current density (the main kinetic parameter, defined as the extrapolated value at zero current or equilibrium state, which thereby has thermodynamic meaning in the broader significance of the Nernst definition of equilibrium electrode potential) the enthalpy of adsorption of intermediates in the rate determining step (RDS) and the work function. Such a particular linear dependence shows that along the energy axis log jo does not follow volcano plots but rather a straight line dependence.14,17,18 Meanwhile, volcano plots usually arise along the periodic table, as the effect of the electronic configuration of ordered elements and the overall periodicity in the majority of physical and chemical properties.10-13 Brewer’s high temperature thermodynamic theory predicts and has experimentally confirmed that whenever one hypo-dmetal alloys or interacts with hyper-d-electronic semiseries of transition elements, and vice versa, typical volcano plots arise

with rather pronounced high values of free energy changes for intermetallic bonding.1-4,8-11 Such theoretical interbonding prospects and experimental evidence reflect, and as a consequence, a priori predict the same type of volcanic dependence in electrocatalytic activity along the inexhaustible variety of hypo-hyper-d- or hypo-f-hyper-d-interelectronic phase diagrams of transition elements (Figure 1), the latter so far being proved on many catalytically significant issues.9-11 In other words, it has been shown that each hypo-hyper-d-intermetallic phase diagram behaves both in the bonding effectiveness (free enthalpy changes) and electrocatalytic activity for the HER as the part of the periodic table between two periods to which initial constituents belong.10,11 Such one-to-one parallelism in the interdependence between the peaks of various Zr-Ni phase stability values and electrocatalytic activity (either the exchange current density or the actual current density at constant overpotential) along the phase diagram correlates the Brewer intermetallic bonding theory and the present model of electrocatalysis for the HELR (Figure 1) in a straightforward, linear manner. The point is that the d-band is both the bonding and adsorptive orbital,9-11 while there exists a simple linear (approximately “six by one”) interdependence of fundamental significance between the enthalpy of bonding (or cohesive energy) and free surface energy,19 that is revealed to be straightforward from the volcano plots along the periodic table.10,20,21 Thus, the stronger the bonding, the weaker the intermediate H-adatom enthalpy of adsorption and, consequently (Sabatier22 principle in catalysis), the faster the hydrogen electrode reactions (HELRs), both cathodic evolution (HER) and its anodic oxidation (HOR). Such a statement has been an a priori logical hypothetical definition, which has then been broadly experimentally proved.11,16 There are lots of various significant consequences of the hypohyper-d-interelectronic interaction of transition metals. The majority of hypo-d-metals, for example, cannot be cathodically deposited from aqueous media. This holds typically for Mo and W, while their co-deposition with hyper-d-elements proceeds both with a rather high final percentage of hypo-d-metals in the resulting alloys and remarkable Faradaic yields.23,24 Since according to the present theory, the underpotential deposition (UPD) of one metal upon another occurs whenever

3032 J. Phys. Chem. B, Vol. 110, No. 7, 2006

Neophytides et al.

their mutual cohesive energy (or bonding effectiveness) exceeds the individual values of the constituents, hypo-hyper-dinterelectronic combinations are the most predestined for such interactive relations. Such a hypo-hyper-d-intermetallic UPD affords the basis for a new generation of submonolayer electrocatalysts.16,25,26 As an example of the effect of interorbital d-band bonding effectiveness, the submonolayer of hypo-hyperd-intermetallic combinations usually leads to a dramatically reduced temperature of CO desorption.27 1.2. Thermodynamic Basis of Facilitated Thermal HypoHyper-d-Intermetallic Phase Formation. Hypo-d-electronic transition metals are rather oxophilic elements and thereby the subject of specific and tedious metallurgical production, while their affinity for intermetallic bonding with hyper-d-metals is pronouncedly high and usually results even in extraordinarily stable Laves type intermetallic phases.1-4,9,10 These are accompanied by unusually high (negative) enthalpies of formation and, as a consequence, present dramatically increased melting points, and rather pronounced synergism in the electrocatalysis for the HER.8-11 The main breakthrough approach of the present method of composite catalyst formation primarily consists of a proper application of the extended Brewer interionic bonding theory,16,25 the use of the hypo-hyper-d-interionic d-d-bonding effect at the molecular level of properly mixed metal ions and/or molecules, for example, their oxides or hydroxides in a consistent stoichiometric ratio, and reduction of the overall reaction on the “neutralization” or water production (steam withdrawal), such as

TiO2 + 3Pt + 2H2 ) TiPt3 + 2H2O

(2)

One of such examples is the interactive decomposition of hyperd-metal acethylacetonates (2,4-pentanedionates, M-acac), along with the spontaneous catalytic reduction of hypo-d-oxides or hydroxides mixed together at their molecular level by a proper sol-gel procedure, such as

Hf(OH)4 + 3Pd(CH3-CH-CO-CH-CH3)2 + H2 f HfPd3 + 8CO + 2CO2 + 20CH4 + 4H2O (3) where at a rather low temperature decomposed Pd-acac results in finely dispersed Pd atoms, as scattered self-catalytic centers (sites) for further interactive bonding into an extrastable Laves phase (HfPd3) under relatively mild conditions. This principle applies in general to any other hyper-d-electronic element participating in equivalent reactions, described by eq 3. In other words, while there are no real technological conditions for reducing individually titania, hafnia, zirconia, ceria, lanthania, and so forth, at reasonably and attainably high temperatures to their metallic state,28 the extended Brewer d-dinteractive method enables one to combine them properly and substantiate the formation of desirable intermetallic phases with the simultaneous reaction with Pt, Pd, Ir, Ni, Co, or their oxides (eq 2) and/or other salts (preferably nitrates, acetates, oxalates, and similar combinations, because of their simple thermal decomposition), at only a few hundred degrees centigrade. In other words, the extraordinarily high enthalpies of formation of the Brewer type stable Laves phases, as a thermodynamic driving force, succeed to a large extent in subtracting the heats necessary to reduce the hypo-d- and/or hypo-f-oxides and hydroxides in such a properly composed overall reaction. The main theoretical impact to establish a novel nanotechnological field of hypo-hyper-d-interionic and intermetallic catalysts is based on the interactive reduction, in which atoms

TABLE 1: Relevant Thermodynamic Values for Present Calculations (after Kubichewski et al.28) sample

∆H298 (kJ mol-1)

S298 (J mol-1 K-1)

TiO2 (s) (rutile) TiO (s) (monocl.) HfO2 (s) (monocl.) ZrO2 (monocl.) H2 (g) Ti (s) H2O (g)

-944.0 -542.7 -1117.5 -1100.8 0 0 -241.8

50.6 34.7 59.4 50.4 130.6 30.7 188.7

of Pt, Pd, Co, Ni, and/or any other suitable hyper-d-electronic transition metal behave both as the direct reactant to produce defined intermetallic phases (TiPt3, HfNi3, ZrPd3, CeCo5, LaNi5, etc.) and as a self-catalyst to reduce hypo-d- or hypo-f-oxides, like TiO2, HfO2, ZrO2, and CeO2, in the course of their mutual interaction. Thus, the overall reaction (eq 2) can be thermodynamically considered to occur as being split into two steps, consisting of the straight, catalytically facilitated (Pt) titania reduction

TiO2 + 2H2 ) Ti + 2H2O

(2-1)

and further the direct Brewer intermetallic bonding reaction

Ti + 3Pt ) TiPt3

(2-2)

The reliable relevant thermodynamic functions for this type of reaction of Pt with Ti, Zr, and Hf are due to Meschter and Worrell,29,30 while the broad thermodynamic survey of data for oxides is due to Kubaschewski et al.28 (Table 1). Since thermodynamic functions and data are additive within a defined stoichiometry (extended Hess law, Haber-Born cycle), from the existing literature, it follows that

∆G1(TiO2 f TiPt3) ) ∆G1-1 + ∆G1-2 ) -RT ln K1 ) (460 400 - 96.3T) + (-341 833 + 33.51T) (4) The main point and substance is in a huge free enthalpy change with the intermetallic hypo-hyper-d-interelectronic phase formation, that succeeds energetically (or catalytically) to compensate for the TiO2 stability. When the equilibrium constants are introduced in the overall dependence, which practically reflects the ratio of partial pressures of hydrogen and water vapor, with other reactants being in their standard states, a parametric relation follows

∆G1(TiO2 f TiPt3) ) 118 567 - 62.79T ) 38.29T log

( ) pH2

pH2O

(5)

When the latter is compared with the individual thermodynamic equation for TiO2 reduction,28

∆G(TiO2) ) 460 400 - 96.3T ) 38.29T log

( ) pH2

pH2O

(6)

it is possible now to compare the temperature (T) of the straight individual titania reduction and the overall Brewer effect of mutual interactive reduction for three such systems, TiPt3, ZrPt3, and HfPt3, along the partial hydrogen versus water vapor pressure relation, taken as an axis (Figure 2). While titania itself thermodynamically requires, for example, a minimum of 2180 K to be directly reduced under simple technical conditions (pH2/ pH2O ) 1000), under the same circumstances, the formation of



Figure 2. Temperature dependence of various hypo-d-oxides (TiO, TiO2, ZrO2, HfO2) reduction to their individual metallic state, and catalytic formation of their Laves type stable intermetallic phases (TiPt3, ZrPt3, HfPt3), as a function of parametric ratio of hydrogen and water vapor partial pressure.

TiPt3 occurs at 667 K. Although hydrogen can be supplied with less than 1 ppm of moisture, this graph is restricted at technically accessible real conditions, since in the course of such processes the main accompanying product is water. On the electrochemical Volta scale, just for comparison purposes, to assess the Brewer interactive effect contribution, the two above reactions would thermodynamically occur in their spontaneous directions at 1.1184 and 0.2587 V, respectively, and such a remarkable difference favors the appearance and growth of strongly bonding Laves intermetallic phases. In other words, while rather strong reducing agents (like NaBH4) are absolutely unable to reduce directly titania, hafnia, or zirconia, their stoichiometric molecular mixtures with Pt, Pd, and Ni salts, resulting in Laves type intermetallic phases, are at least partially attainable (accessible). In such a context, it is worthwhile to note that, in agreement with the present theory, the literature shows usual (enough interactive) annealing at about 300-400 °C for all intermetallic catalysts, although this is quite below the individual melting points. The same type of thermodynamic relations for Hf and Zr corresponding to eq 5 are the following:28-30

∆G(HfO2 f HfPt3) ) (633 900 - 100.3T) + (-475 302 + 52.30T) ) 158 598 - 48.0T (J mol-1) ) pH2 (7) 38.29T log pH2O

( )

∆G(ZrO2 f ZrPt3) ) (617 200 - 104.8T) + (-452 374 + 52.72T) ) 164 826 - 52.08T (J mol-1) ) pH2 (8) 38.29T log pH2O

( )

This is the substance and basis for a novel nanotechnology of hypo-hyper-d-intermetallic nanoparticulate cluster phase formation: the stronger the interionic d-d-bonding, the lower the thermodynamic temperature of intermetallic phase formation. The bonding effectiveness apparently increases with the exposure of d-orbitals (Figure 2), from the 3d level toward the 5d level, or from Ti toward Hf, and this confirms the initial basic statements of the Brewer intermetallic bonding theory.1-4 The

broad common point is that all hypo-hyper-d- and/or hypo-fhyper-d-intermetallic phases characterize a strong interionic (or interatomic) bonding effectiveness, and this gives place to the versatile and broad number of combinations by simply following the present technology and thermal procedure. The same thermodynamic calculations based on the lower oxide states of TiO, ZrO, and HfO, as the starting reactants, show that in contact with Pt they are absolutely unstable and instantaneously produce stable TiPt3, ZrPt3, and HfPt3 intermetallic phases. Such a fact is noteworthy, since supported catalysts (M/TiO2) immediately undergo a rather facilitated reduction of titania by hydrogen in a furnace or even by adsorbed H-adatoms (M-H) during heterogeneous catalytic processes, to a nonstoichiometric suboxide state (TiOx, where 2 > x > 1), that finally results either in Magneli phases (TinO2n-1)31 or in defined intermetallic phases. The choice of Ti, Zr, and Hf as hypo-d-components in the intermetallic bonding reaction with selected hyper-d-electronic transition metals (Pt, Pd, Ru, Rh, Ni, Co) has been based on the following decisive facts: (a) Brewer intermetallic bonding theory a priori predicts rather strong ionic interactions, accompanied by corresponding rather pronounced thermal effects, and thereby a rather high stability of the Laves phases1-4 formed, and consequently, in accordance with the present theory, rather high synergistic electrocatalytic effects for the HELR.8-11 (b) Such rather pronounced oxophilic features of Ti, Zr, and Hf,32 in combination with Pt and similar metals, enable the peaks of primary oxide formation (Pt-OH) in their common potentiodynamic spectra to be shifted toward much less positive potential values, ideally to coincide with hydrogen desorption peaks, and thereby to provide almost simultaneous anodic oxidation of CO and H-adatoms, and remarkably increase the so-called “CO tolerance” in low temperature PEM fuel cell (LT PEM FC)16,33 catalysts. Such oxophilic contributions can be obtained either straightforwardly in direct specific intermetallic phase compositions, for example, RuPt, MoPt4, and PtSn, or by the primary oxide (M-OH) spillover from oxophilic hypod-oxide supports. In other words, Ti, Zr, and Hf are supposed to take their more pronounced oxophilic role in the new generation of electrocatalysts proposed in the present paper, than Mo, W, Ru, and/or Sn themselves. In supported AB3/TiO2 or AB3/TinO2n-1 type catalysts, which implies wet hypo-d-oxide supports, the spillover of primary oxides (M-OH) should be rather more pronounced.16,25,33 1.3. Strong Metal-Support Interaction, Its Concepts, Nature, and Consequences. The whole contemporary heterogeneous catalysis is based on supported catalysts, where individual or prevailingly hyper-d-electronic metal arises in the strong metal-support interaction (SMSI) with mostly hypo-doxide (TiO2, ZrO2, HfO2) or various hypo-d-hypo-f-oxide supporting composites (TiO2, CeO2).34-39 Brewer1-4 hypohyper-d-interelectronic bonding theory precedes and in its extended interionic bonding model16,25 a priori anticipates Tauster34-36 SMSI, and with such interactive properties has recently been the basis for catalyst grafting (anchoring) upon suitable hypo-d-oxide supports.16,25 The operational basic Tauster34-36 definition of SMSI implies that the metal-oxide interaction (M/TiO2) for heterogeneous catalysis in the gas phase, in accordance with the bonding strength, results in substantial weakening and even suppression of both M-H and M-CO intermediate chemisorptive bonds. Meanwhile, new concepts additionally imply the spillover of interactive primary oxides (M-OH)16,25,33 and their decisive interference in the

3034 J. Phys. Chem. B, Vol. 110, No. 7, 2006 catalytic process. In general, the stronger the intermetallic hypohyper-d-interatomic bonding, the weaker the adsorptive bonding of intermediates in the RDS, and consequently the higher electrocatalytic activity for the HELR.8-11 Following the same rule, the stronger the hypo-hyper-d-interionic metal-oxide support interaction, the higher the catalytic activity.16,25,33 The SMSI has been observed only on those metal systems that spontaneously and adsorptively dissociate hydrogen and enable its spillover interaction with the support,34-39 with the exception being strongly bonded Au (SMSI) on titania. In such a respect, there is a strong indication of the substantial similarity between the electronic interactions taking place within a composite hypohyper-d-intermetallic phase and those at the strongly reduced TiO2 supported metal catalysts, albeit a chemical bond type of bond has usually been ascribed to them.39 In other words, contemporary heterogeneous catalysis already implies that the electronic interactions in SMSI are similar to those in the Brewer type intermetallic phases.40,41 Chemisorption studies indeed show that the otherwise irreVersible adsorption of CO no longer occurs as such on highly dispersed Pt particles interactively bonded to anatase titania (TiO2).37-39 In addition, under CO hydrogenation reaction conditions, the IR bands characteristic of CO irreversible chemisorptive bonds could no longer be detected on Pt/ TiO2 catalysts, while they were still present on non-hypo-delectronic oxide supports (SiO2, Al2O3).37-39 In fact, for metal atoms at the interface or directly contiguous to the TiO2 support around the nanostructured circumference, their common intermetallics grow by the facility of titania reduction. According to such a hypothesis, the catalytic activity should increase with the interfacial perimeter, which, in its turn, is roughly proportional to the second power of the metal dispersion. As a concluding remark on catalysis, the ability of some hypo-delectronic oxide supports of transition metals (such as anatase titania, TiO2; ceria, CeO2; zirconia, ZrO2; hafnia, HfO2) to undergo reduction and form the intermetallic phase enables the Brewer type strong interactive bonding at the interphase. The latter mostly and substantially results in a priori defined symmetric Laves type intermetallic phases, otherwise characterized by the maximal free enthalpy of formation (TiPt3, TiPd3, TiNi3, CePt5, HfPt3, ZrNi3, etc.), just in accordance with the above thermodynamic theoretical consideration, and this lies in the core of the SMSI, and arises as a naturally implied issue of the extended Brewer interionic bonding theory.16,25 While neither Au nor titania by itself is individually active for CO and O2 adsorption, nanostructured Au/TiO2 oxidizes CO even well below room temperature.42-49 In light of the existing electrocatalytic theory,8-11 such an effect could a priori be expected, if the original Brewer interatomic bonding theory, with all accompanying consequences,1-4 holds in its extended concept for the interactive interionic bonding.11,16,25 In other words, massive Au and nanostructured supported Au/TiO2, or Pt and Pt/TiO2, in their catalytic features are distinctly different species. An abundance of papers,40 books,37,38 and chapters39 in heterogeneous catalysis have been based on, and following the original idea of Tauster,34-36 defining and applying the properties and consequences of SMSI, while there have so far been no theoretical explanations concerning the driVing force, causes, and nature of such an interactiWe bonding. The core and quintessence of such a strong metal-support interaction almost without any exception lies in the Brewer hypo-hyper-d-interelectronic bonding effect resulting in interactive properties. In other words, two interrelating effects should be distinguished: (a) the strong hypo-hyper-d-d-metal-support bonding and (b) the interactive primary oxide (M-OH) spillover,16,25 as the dynamic

Neophytides et al. electrocatalytic effect, or other “decoration” (or “encapsulation”37-39) features arising from the partial hypo-d-oxide support reduction. This is the main aim and most significant theoretical statement and contribution of the present paper. The d-orbital is the prevailing bonding and chemisorptive band1-7,10,11,19,20 and therefore the decisive catalytic and electrocatalytic band,10,11,16,25 while the s,p-band contributes a constant bonding term all along each transition series.1-4,50 The higher the exposure of the d-orbital in space, the higher both the individual bonding effectiveness and catalytic activity, so that an increase from the 3d level to the 5d level8-11 results in typical volcano plots for d-band energy, and consequently, Fermi, cohesive, and free surface energy1-4,8-11,20 are linearly interrelated. Such a conclusion agrees well with the statement of Haruta42-49 that Au with almost filled d-orbitals, at a smaller particle size than 2 nm, in particular when interactively supported on hypo-d-oxides, due to its additional forced or strained d-orbital exposure, imposes a strikingly different catalytic behavior for many catalytic reactions, including enhanced and superior CO oxidation. Goodman et al.51 named such an effect the quantum size effect (QSE). Mavrikakis et al.52 use the step density, defined as the fraction of atoms in cluster particles having a smaller number of neighbors with decreasing nanosize, that exponentially grows with a decrease of the total number of atoms in such cluster units, and thereby correlates in a straightforward manner with catalytic activity; the cause is the samesthe strained (or forced, according to Haruta42-49) exposure of d-orbitals within a smaller cluster size occurring in particular at crystal edge steps. In other words, such a structural cluster architecture with a decreasing number of neighbors implies more expanded interbonding d-orbitals, that is even more pronounced at the interphase SMSI contact with a hypo-d-oxide support. The same cause for the extended d-band exposure in space, as already pointed out,8-11 certainly arises in the hypo-hyper-d-interelectronic interaction of the Brewer type intermetallic Laves phases and stays in the core of the synergistic advances in electrocatalysis of the HELR.16,25 The recent approach in the density functional theory52-54 uses the d-band center (d) as a parameter which measures the energy of the metal d-states relative to the Fermi energy, to establish the energy of segregation of alloying transition metals and therefrom the adsorption of H-adatoms52 or CO.53,54 The conclusions are quite the same as those in the present theory, since segregation is a reciprocal function of cohesive d-dbonding effectiveness and/or the d-band energy: chemisorption of H-adatoms decreases with decreasing segregation energy and in the same relation increases the electrocatalytic activity for the HELR.52 Such theory predicts that prevailing hyper-d-metal, as a host, pulls inside the phase the hypo-d-constituent of such an alloying composition and the attracting force depends on their hypo-hyper-d-intermetallic bonding affinity.53,54 Interactions within two transition metal d-bands are metal-dependent and treated as perturbation.52 Qualitatively, there is no difference with the present theory statements. However, one thing is the hypo-hyper-d-intermetallic bonding and the resulting adsorption of H-adatoms, and quite another when the species in contact are oxygen and water molecules. The latter implies another type of strong interaction with otherwise oxophilic hypo-d-ingredients and imposing attraction to pull them back in the opposite direction, to segregate as oxides upon the surface. The highest oxide valence states block active catalytic centers and decrease the overall catalytic activity, while the lower oxide states usually exhibit electronic conductivity, rather high catalytic activity, and spillover of the primary oxides (M-OH).33

Supported Interactive Electrocatalysts The present paper aim and purpose is to employ conductive and SMSI Magneli phases and anatase titania submonolayer modified Magneli phases, sometimes even advanced in the altervalent capacity by partial replacement of TiO2 with tungstenia (WO3), as interactive and highly electronic conductive electrocatalyst supports. The interbonding hypo-hyper-dinterelectronic effect, along with the interactive spillover of the primary oxides (M-OH), as a dynamic catalytic effect, is theoretically predicted and expected to increase the electrocatalytic activity and the intermetallic synergism for the HELR, as well as dramatically improve the conductivity and, consequently, the current collection and integration of nanostructured catalytic particles, primarily in the overall contribution to the efficiency of LT PEM FCs and water electrolyzers. 2.0. Experimental Section The majority of catalytic materials were prepared by solgel synthesis. First of all, anatase titania as the main catalytic support was produced by hydrolysis of titanium isopropoxide in anhydrous ethanol and spread as a submonolayer upon Vulcan XC-72 carbon or Magneli phases, and after supercritical drying with CO2, it was calcinated in air at 300-400 °C. The sol-gel process to obtain 10 nm sized peroxopolytungstic acid (PTA) sol was adapted following the Kudo,57 Livage,58 and Orel59 procedures, which start with dissolved small fine particles of metallic W in 30 wt % H2O2, followed by decomposition of the excess of the latter by the catalytic effect of a piece of Pt. Controlled and repeated evaporation was conducted from onethird to one-fourth of the initial aqueous volume in a mixture containing 5 times the amount of anhydrous ethanol at a carefully fixed temperature below 70 °C, where mostly an azeotropic mixture leaves the phase. Such an ethanol solution of homogeneous colloidal particles (size of about 10 nm) of PTA was then mixed with titanium isopropoxide to get a polymeric network of about 3-7 mol % PTA, calculated on the basis of WO3, that was the main composite hypo-d-electronic oxide support for SMSI. In this way, the altervalent capacity of titania (z ) 4) was increased to a higher value (z ) 6), for much advanced primary oxide (M-OH) spillover. Magneli phases of nonstoichiometric titania or titanium suboxide (general composition TinO2n-1, 10 g n g 4, on average Ti4O7, Ebonex, Atraveda Ltd., Mansfield, U.K.), with advanced conductivity (300-1000 S cm-1), a particle size below 10 µm, or grinded in planetary mills to 1 order of magnitude smaller size, were the main catalytic support for the very first time introduced in electrocatalysis, either plain or with a submonolayer of anatase titania (TiO2) and/or in an admixture of the latter with PTA. Magneli phases are the most advantageous both current collecting and conducting support but are of much smaller available surface area (0.2-1.2 m2 g-1) relative to anatase (∼250 m2 g-1), and as it is shown in the present study, the hypo-d-suboxide has good interactive spillover properties in the primary oxide transfer. The main and advantageous solvent for providing a homogeneous dispersion of individual colloidal catalytic precursors in a rather simplified manner, and/or properly mixed hypohyper-d-interionic composites, was the buffered (pH > 12) alkaline (NaOH, Na2CO3) ethylene glycol (EG).55,56 Glycol and/ or propylene glycol can also be used, but the dielectric constant is mostly favorable for EG. The point is that, at lower pH values, below the isoelectric point, agglomeration leads to larger nanosizes of the resulting metallic clusters, while, at pH > 12, there a rather uniform distribution of about 2 nm appears.60 The point is that the pzc position, as located in the acidic pH range

J. Phys. Chem. B, Vol. 110, No. 7, 2006 3035 SCHEME 1: Model of Interactive Grafting of Pt-acac (Acethylacetonate, Pt-2,4-pentaneditionate) upon Titania Support

(pH 4-6), provides a positive charge distribution upon anatase titania and over Magneli phase surfaces in strong alkali media, this way enabling a suitable grafting interaction with negatively charged colloidal particles or individual composite catalytic precursors (Scheme 1). Various individual or composite mixtures of transition metal acethylacetonates (in short, M-acac or M-2,4-pentanedionates) were used as precursors to produce single or intermetallic phase nanoparticles on the above-stated catalytic hypo-d-oxide supports. The advantages are their low temperature decomposition, even unnecessary pH control, but preferably alkaline media if in aqueous solution, and the absence of any ionic species accompanying the resulting metallic clusters while in ethanol or 2-propanol and even acethylacetonate itself. However, much cheaper common chemicals for intermetallic phase electrocatalyst production can be used, like the various available salts such as hexachloroplatinic acid, Pd(NO3)2, ammonium molybdate, and tungstenate, but these require proper filtration and washing to remove the remaining ionic species. As a rather strong reducing agent, EG enables noble metal clusters to be obtained quantitatively and evenly grafted on the exposed surface of hypo-d-oxide support surfaces simply at higher (about 160 °C) temperatures and during a relatively short period (4-6 h) of time.55,56 However, intermetallic phases including more oxophilic hypo-d-metals require in addition some stronger reducing chemicals (HCHO, NaBH4), even when dealing with M-acac precursors. Although it is not always mandatory, all the supported nanostructured electrocatalysts produced in this way were thermally treated and annealed in a hydrogen furnace for at least 4 h usually at 400 °C, particularly when dealing with a higher degree of oxophilicity components (TiO2, ZrO2, HfO2). For noble metals, a calcination in air rather than in a hydrogen stream is preferable, resulting in a more developed surface area. Titanium(IV)-diisopropoxide-bis(2,4-pentanedionate), zirconium oxynitrate (ZrO(NO3)2), and Hf-acac {Hf(CH3-CO-CHCO-CH3)2} were used as the source of reacting Ti, Zr, and Hf, respectively, while Ni and Co, either as nitrate salts or acethylacetonates. Some simple combinations of various hypo-hyper-d-intermetallic M-acac precursors were dissolved in ethanol and mixed at their molecular level for the desired intermetallic ratio, leading to an a priori specified intermetallic phase, sometimes with a small amount of 2,4-pentandionate to facilitate their complete dissolution. The latter were grafted upon hypo-d-oxide supports, evaporated, and subjected to reductive thermal decomposition in a hydrogen furnace, or in the case of extreme oxophilic compounds to direct reduction by formaldehyde or preferably NaBH4 in EG to keep a fine colloidal dispersion for later nanostructured intermetallic phase growth in a hydrogen stream. In all cases, traces of stimulating reagents for proper crystal-


Figure 3. XPS Pt 4f spectra scanned at the interphase between Pt catalyst (1.0 wt %) and anatase TiO2 support in their SMSI. Deconvolution exactly reveals the existence of the TiPt3 intermetallic phase with all identified spectral properties.

lization in the desired face-centered cubic (fcc) system of nanostructured clusters, that naturally corresponds to prevailing hyper-d-electronic components, such as Cd or Pb salts (in ∼10-4 M content), were employed. Amorphous nanoparticles never provide a synergistic electrocatalytic effect, but only different crystal structures from the natural crystal system for every given hypo-hyper-d-interionic composition, which in particular for the HELR is fcc,1-4,8-11 as was also recently confirmed by Haruta42-49 and Mavrikakis et al.50,52 Electrocatalysts obtained by reduction of individual or prevailingly hyper-d-electronic metals (in their combinations with oxophilic hypo-d-components), that adsorb hydrogen from the gas phase in its furnace, both being supported on hypo-d-oxides, during such a thermal treatment undergo a partial transformation of anatase titania into Magneli phases, so that X-ray photoelectron spectroscopy (XPS) analysis shows no charge separation. Other details of experimental procedures, in particular concerning the apparatus and methods of measurements, were described in our preceding papers.16,25,26,33 3.0. Results, Discussion, and Consequences 3.1. Thermal Hydrogenation of Titania into TiPt3 Intermetallic Bonding and Interactive Phase. The inspiring impact and impetus to establish a novel nanotechnology and a new generation of supported composite catalysts and electrocatalysts based on more electronegative and more oxophilic hypo-dcomponents, Ti, Zr, Hf, rather than Mo, W, V, Cr, came from an important observation. During typical thermal production of supported catalysts, in a hydrogen furnace, at the interphase Pt/TiO2, and in accordance with the present thermodynamically based theory, a well-defined and XPS confirmed (Figure 3) intermetallic TiPt3 phase arises at rather low temperatures (250350 °C). This phase, as already intuitively predicted by Tauster,34-36 arises because of the SMSI and its further interactive catalytic effects.16,25,26 The whole thermodynamic approach displayed above is based and calculated on equilibria with molecular hydrogen, while in the present issue H-adatoms (Pt-H) present an even higher activity and facilitate the overall reduction of TiO2, the effect known as encapsulation or decoration,37-39 finally resulting in its stable and interactive intermetallic (TiPt3) phase. In other words, as inferred above, such reduction steadily leads to titania suboxide (TiOx, x < 2), and when stoichiometry approaches a value of unity (x f 1),

Neophytides et al. in the presence of Pt, the thermodynamically anticipated most stable intermetallic phase, TiPt3, instantaneously appears. In fact, it is easy to show that such an intermetallic phase between the metallic catalyst and its hypo-d-oxide support imposes the overall SMSI effect, and thereby governs the entire catalytic process and gives all of the main interactive properties of supported catalysts.16,25 The more intense Pt 4f7/2 component centered at 71.6 ( 0.1 eV is shifted by 0.6 ( 0.1 eV to higher binding energies compared with a clean Pt surface (Figure 3). Such an energy shift is consistent and reproducible in all measurements done before, after and during annealing of the sample at 570 K, and corresponds to Pt ions of the Laves type Brewer TiPt3 intermetallic phase,1-4 in agreement with literature values,61,62 in particular since a distinct enhancement in intensity of the peak ascribed to TiPt3 arises after annealing. This clearly shows that the final state effects have restricted the influence on the binding energy shift of the measured Pt 4f peak, indicating that Pt particles are not in nanoscopic dimensions as to give the so-called quantum size effect in X-ray photoelectron spectroscopy. Such a state of experimental evidence strongly indicates the existence of the TiPt3 Laves phase within the interphase between metallic Pt and anatase titania (Pt/TiO2) after proper heating in a hydrogen stream. In such a respect, the heating time under the adsorptive state of H-adatoms on Pt, as a parameter of interionic interaction, is even more important to produce such an intermetallic phase (TiPt3) than the temperature level itself. The presence of the TiPt3 phase, at a rather dispersed Pt on the titania surface, testifies for the rather strong interionic bonding between Pt and Ti (∆ ≈ 1.0 eV). Among all possible intermetallic phases along the Ti-Pt phase diagram, the affinity and thermodynamic properties a priori indicate the appearance of TiPt3 as the most strongly bonded and most stable species1-4 (Figure 3). In fact, the latter (TiPt3) marks the main interbonding and interactive characteristics for SMSI as being decisive for further electrochemical behavior and defines the main catalytic properties of such composite electrocatalysts. At the same time, such experimental insight affords the best illustrative picture and reveals the deeper physical meaning and nature of the hypohyper-d-interelectronic bonding effectiveness, and thereby of the interionic grafting abilities, together with the resulting physicochemical consequences in electrocatalysis. Finally, it confirms the above-displayed thermodynamic theory for the thermal production of such Brewer type electrocatalysts. 3.2. Magneli Phase Interactive Supported Pt as Advanced Electrocatalyst for Oxygen Reduction (ORR). The most outstanding electrocatalytic effect with interactive Magneli phase support has recently been substantiated by the extremely highly dispersed Pt sub-nanostructured monatomic particles. Such a Pt deposit of rather unusual size, as high as 36 m2 g-1 Pt upon a rather small (1.6 m2 g-1) available Magneli phase specific area, was obtained from a 2-propanol solution of Pt(NH3)2(NO2)2 by interactive grafting deposition followed by reduction in a hydrogen stream at 300 °C.63 High resolution transmission electron microscopy (TEM) analysis (Figure 4) shows a particle distribution between 0.1 and 1.0 nm, with prevailing monatomic dispersion, though on average only every eighth Pt atom is directly hypo-hyper-d-interelectronically bonded on surface Ti atoms of Magneli phases. A smaller particle size in metal deposition so far has not been TEM recorded. Polarization properties were exactly experimentally assessed on the basis of the estimated surface area of equivalent UPD H-adatoms and compared with so far the best Pt/Vulcan-XC-72 electrocatalyst



Figure 6. Charge density that required oxygen species for their reduction, presented as a function of potential for the Pt/(Magneli phases) (5.4 µg Pt) electrode. The inset shows potentiodynamic I vs E relations scanned from different initial potentials (hold) with a sweep rate of 5 V s-1.

Figure 4. TEM images of Pt nanoparticles monatomically dispersed upon Magneli phases (Ebonex, Atraverda) for low and high magnification.

Figure 5. Tafel plots for the cathodic ORR scanned on monatomic dispersed Pt (5.4 µm cm-2) upon Magneli phases (open circuits) and nanostructured Pt (10 wt % Pt on Vulcan XC-72, triangles) in 0.5 M HClO4 solution.

(E-TEK, Somerset, NJ, 10 wt % Pt) for the ORR (Figure 5). Kinetically controlled current density values at a potential of 0.85 V versus RHE, where the mass transfer effect is negligible, were 0.33 mA cm-2 Pt for pure Pt/C and 0.61 mA cm-2 Pt for Magneli phase supported Pt (5.4 µg cm-2). With about 2 orders of magnitude increased Magneli phase specific surface area and one-to-one bonded Pt-Ti atoms, such outstanding results could be further advanced. Three aspects of such remarkable electrocatalytic improvements can be accounted for by the following: (a) A monatomic network dispersion implies a dramatically increased electronic density of states along with the highest d-orbital strain effect, and consequently advanced catalytic activity. (b) The SMSI effect exponentially increases52 with decreasing metallic catalyst

particle size and approaches its maximum for monatomic dispersion. (c) The interactive Pt-OH spillover effect, which cannot be separately and unequivocally estimated in the presence of the other two effects in aqueous media and for Magneli phases is doubtless present and contributes to the overall catalytic properties. To check the value of the eventual coverage with oxygen containing species in the range of potential close to the open circuit value (0.9 V for Pt/C and 1.03 V for Pt/(Magneli phases) vs RHE) for a Pt/Ebonex (5.4 µg Pt) electrode, some specific additional potentiodynamic measurements were carried out. The electrode, previously held at 0.20 V (RHE), was potentiostated for 30 s at the selected initial potential in the range characterized by the lower Tafel slope on polycrystalline Pt (Figure 5). In the next step, the potential was linearly scanned with a rather fast sweep rate of 5 V s-1 down to 0.0 V (RHE) (Figure 6, inset). Such scanning circumstances enable the appearance of peaks for slow reaction steps to be avoided. The amount of charge (Figure 6) spread within characteristic peaks during the scanning period, determined by integration of I-E curves, proved that the surface coverage by Pt-OH on thePt/ (Magneli phases) (5.4 µg Pt) electrode was negligible and the adsorption conditions were Langmuirian. The absence of the Pt-OH layer, that eventually might inhibit oxygen reduction on polycrystalline Pt, is explained by the synergistic effect upon the catalyst, as a consequence of the interactions of Pt nanoparticles with Magneli phases, resulting in the spillover of Pt-OH and its spontaneous cathodic reduction by the fast reversible electrode reaction (eq 9). In other words, the fast reversible electrode reaction which proceeds independently within a critical potential range can by no means contribute to the accumulation of the main reacting species (Pt-OH) upon the Magneli phase supported Pt electrode surface regardless of the fact that the latter appears and takes place in the whole mechanism of the ORR. 3.3. SMSI of Gold upon Anatase Titania. Haruta et al.64 have shown that the same reactants (propylene along with an equimolar mixture of hydrogen and oxygen) yield different products upon different nanosized Au catalysts supported on anatase titania (Au/TiO2): (i) propane by hydrogenation at nanoparticles 2 nm Au. Hydrogenation implies H-adatom adsorption upon Au, that never occurs on a pure


Neophytides et al.

Figure 7. XP spectra of Au 4f for vapor deposited nanostructured Au upon a fine thin film of anatase titania with deconvolution for lower amounts of deposits to reveal the existence of primary oxides (Au-OH and AuOOH).65

massive gold surface. Haruta42-45,47-49 ascribes such chemisorptive properties to “forced Au-d-orbitals”, in particular because of the SMSI when deposited on anatase titania. In such a respect, rather fine Au films were investigated when deposited by electron beam evaporation from a Au wire of ultrahigh purity under vacuum higher than 6 × 10-4 Pa, onto stationary titania coated microscopical slides, placed in parallel to the emitting surface.65 Rough and high-surface-area nanocrystallyne titania thin films for Au vacuum deposition were prepared on microscopy glass slides by applying the doctor-blade procedure usually employed for photocatalytic substrates.66 A quartz crystal monitor was used for controlling the deposition rate of 0.01 µg cm-2 s-1, and by varying the deposition time, the following Au deposits were obtained: 0.4, 0.8, 1.6, and 2.0 µg cm-2. The XP spectra of the Au 4f electrons reveal the remarkable binding shift (Figure 7), that testifies for the SMSI on the interphase Au/TiO2, and this is one of the first evidences of such an effect in heterogeneous catalysis.67 The smaller the nanoparticle size of the Au deposit, as theoretically predicted in the present paper, the larger the binding energy shift with titania, and the more pronounced the whole SMSI. Such a Mavrikakis52 type of trend in the bonding effectiveness and consequently the higher Haruta42-45 type catalytic activity lead us to develop a reversible hydrogen electrode on highly dispersed sub-nanostructured Au SMSI upon anatase titania and/or Magneli phases.68 The whole rather pronounced electrocatalytic and reversible electrode properties result, like the monatomic Pt upon Magneli phases, in the effect of its strained d-orbitals due to a rather high dispersion of nanoparticulate Au and the strong interactive bonding effectiveness (SMSI) upon the hypo-d-electronic support. Such a state of experimental evidence stays in the core of the oxide supported Au behavior as the reversible hydrogen electrode, otherwise unimaginable for its bulky metallic status. The deconvoluted Au 4f peaks with lower Au loadings reveal that Au nanoparticles in interactive bonding contact with titania

are partially oxidized.65 The peak located at 82.15 ( 0.1 eV is identified and attributed to metallic Au, while the peak at 84.05 ( 0.1 eV is situated at 1.9 eV toward higher binding energy and corresponds to the primary (Au-OH or AuOOH) oxides. The latter, in accordance with the present theory, appear as the spillover effect associated with anatase titania and are a priori available for anodic CO oxidation. 3.4. Strong Supported Bonded Electrocatalysts, Their Interactive Properties, and Prospects. Transition metal ions feature several valence states, giving rise to mixed valence compounds (TiO2/WO3), so that such hydrous oxide networks, in particular of polyvalent hypo-d-electronic elements, substantially behave as ion exchange membranes.16,25,26,58 In fact, gels (aero and xerogels) are biphasic systems in which solvent (water) molecules are trapped inside an oxide network, and such material can be considered as a water-oxide composite.25,58 Water molecules are adsorbed at the surface of the oxide particles within such a supporting oxide network, and consequently, such a composite material exhibits specific properties arising from the intimate mixing of both phases. There is a strong firstprinciples thermodynamic confirmation (density functional calculations, DFC)69-71 that water molecules undergo spontaneous dissociative adsorption on anatase and even rutile titania surfaces. More specifically, the (101) surface of anatase characterizes molecular adsorption with partial dissociation (less than 50%), while at the (001) surface proceeds a spontaneous and mainly dissociative chemisorption of water molecules (Figure 8).69,70 In addition, there has recently been shown by the first-principles molecular-dynamic simulations the existence of a mechanism for thermodynamically favored spontaneous dissociation of water at low coverage at oxygen vacancies of the anatase (101) surface with respect to molecular adsorption.71 This is the status of reversible open circuit dissociative adsorption of water molecules at the equilibrium state, while the presence of the metallic SMSI part of catalyst, directional



electric field (or electrode polarization) further disturbs such an established equilibrium and dynamically imposes further continuous dissociation of water molecules. Such a favorable structure of anatase titania for the spontaneously dissociative adsorption of water molecules affords the proper interactive substrate for the grafting adsorption of individual and prevailing hyper-d-electronic nanostructured precursors (Scheme 1). The adsorbing substrate features positively charged properties versus pzc, while the grafting species has a negative overall colloidal charge. In conjunction with such a property, it would be worthwhile mentioning that properly oxidized carbon supports, on which surface appear orderly and homogeneously distributed interactive C-OH groups,72 behave as centers for the grafting of colloidal catalytic precursors, in the same way as discussed above, and display the spontaneously dissociatively adsorbed OH groups on anatase or Magneli phases themselves. The interactive features between hypo-d-oxide, both stoichiometric anatase titania and nonstoichiometric Magneli suboxide phase supports, and water impose several decisive and mutually interrelating effects for the overall catalytic properties of supported electrocatalysts. First, the pronounced effect of moisture43-46 for the entire behavior of titania can be estimated and quantitatively assessed from the dependence of the thermodynamic temperature of TiO and TiO2 reduction in a hydrogen stream with an extremely small amount of water molecules. The variation of the water ratio (H2O/H2) from 10-6 to 10-10 enables the temperature of TiO reduction to decrease from 2020 to 1250 K.35 The main consequences of substantial significance are (a) continuous undisturbed reversible altervalent changes (Ti3+ S Ti4+):

Ti(OH)4 + M f Ti(OH)3+ + M-OH + e-

(9a)

Ti(OH)3+ + 2H2O f Ti(OH)4 + H3O+

(9b)

Σ M + 2H2O f M-OH + H3O+ + e-

(9)

resulting in OH- transfer within such an ion exchange membrane, with consequent spillover of primary oxide (M-OH), over metallic catalyst particles16,25 (Scheme 2), and (b) the spontaneous reduction of TiO2 by chemisorbed H-adatoms (MH) to its suboxide state (TiOx, 2 > x > 1), the latter leading at higher temperatures either to the formation of stable intermetallic phases (TiPt3) at the interphase, as shown above, or to the highly stable nonstoichiometric Magneli phases (Ti2nO2n-1, on average Ti4O7). The above-displayed thermodynamic analysis and the resulting thermodynamic data reflect the initial roots and substantial nature of the moisture effect for hypo-d-oxide support43-46 interactive properties. Thus, as Haruta43-47 experimentally proved, there is no catalytic reaction on supported anatase titania and/or other hypo-d-oxide supports, without moisture, which in anodic CO oxidation relies on the reversible properties of M-OH:

M-OH + M-CO + H2O f CO2 + 2M + H3O+ + e(10) while, in heterogeneous catalysis at higher temperatures, molecular oxygen plays the spillover oxidizing role in electron transfer.43 As a general condition for the semiconducting behavior of transition metal hypo-d-oxides, they appear capable of existing in several altervalent states, exhibiting hopping

Figure 8. Perspective views of DFT-optimized atomic structures for (a) the clean anatase (ADM) ad-molecule model of the unreconstructed (001) surface, (b) the dissociated state of water (0.5 monolayer) on (001), (c) the relaxed geometries of the molecular state of adsorbed water (1.0 monolayer of hydroxylated anatase) on (001), and (d) the mixed state of water on (001) with a half-dissociated coverage of adsorbed monolayer water molecules (courtesy of A. Vittadini69,70).

SCHEME 2: Model Interpretation of the Primary Oxide (Pt-OH) Spillover from Anatase Titania upon the SMSI Pt Catalyst

processes within the solid phase (electrochromic properties16,25,58), the conduction occurs by electron transfer from low to high valence states. Anatase titania features metastable properties and maximal degrees of freedom, while the most compact layerlike and surface bonded Magneli phases show a rather high chemical stability and fascinating electron conductivity, along with the primary oxide transfer due to its suboxide structure.31 Electrocatalytic achievements in the present work introduce and have been based upon interactive hypo-d-oxide supports, primarily as a sol-gel produced submonolayer of anatase titania calcinated on a fine divided carbon (Vulan-XC-72) carrier (TiO2/ C), or its mixed network (3.0-7.0 mol % WO3) with tungstenia.16,25 Such supports, for example, behave spillover of primary (M-OH) oxide responsible for a reversible anodic CO oxidation on Pt nanoparticles, contrary to the non-oxide supported Pt (Pt/ C), so that adsorption and desorption of surface oxides (PtdO) keep the same undisturbed charge capacity within cyclic voltammograms, both in the presence and absence of CO.16 It turned out that nonstoichiometric Magneli phases as catalytic support feature more or less the same order of M-OH transfer (one by four relative to titania) regardless of its stronger interatomic bonding within their lattice, as indicated by the existence of a well delineated peak for its primary oxide


Figure 9. Striping voltammograms for CO desorption from supported 10 wt % (0.4 mg cm-2) RuPt/TiO2/C electrocatalyst CO saturated at three different temperatures: 25 (a), 60 (b), and 80 (c) °C, scanned at a scan rate of 2 mV s-1; (d) the same stripping scans for CO desorption at 60 °C from unsupported 30 wt % (0.5 mg cm-2) RuPt/C electrocatalyst and at a sweep rate of 10 mV s-1; and its CO saturation at 55 °C.

formation along cyclic voltammograms.16 However, this titanium suboxide material as an interactive catalytic choice is advantageous because of its extraordinarily high electron conductivity. Thus, metallic catalysts have been grafted either straight upon Magneli phases or upon a submonolayer covered anatase titania and/or the mixed titania-tungstenia upon the latter, to increase the altervalent capacity from 4 for Ti to 6 for W. This is a novel approach in electrocatalysis, since Magneli phases undertake both the interactive (SMSI) catalyst support features including the M-OH spillover (contrary to the common noninteractive Vulcan XC-72 carbon particles), and at the same time the role of advanced electronic conductor and current collector. The whole effect is much enhanced by the submonolayer of titania upon Magneli phases because of the dramatically increased overall surface area (∼200 m2 g-1), and while the latter is partially reduced to the suboxide state, the whole support keeps a rather pronounced electronic conductivity. In other words, there are either simple bonding and carrying supports, like fine Vulcan XC-72 carbon powder, and interactive ones, usually hypo-d-oxides, like titania, and titanium suboxides, like Magneli phases, with a well pronounced interactive (SMSI) M-OH spillover role. In fact, even the sol-gel produced anatase titania support (TiO2/C) turned out to become partially thermally and catalytically (M/TiOx/C) transferred on a Magneli phase to use at once both their supporting and conductive properties, which in particular occurs on a submonolayer titania modified Magneli phase surface. Ever since Watanabe73 showed that Ru even in a submonolayer deposit, or while alloying with Pt, shifts the primary oxide growth to much more negative potentials and enables CO tolerance, the primary oxide spillover became of substantial significance for LT PEM FCs. Like the hypo-d-oxide supported Pt and Au in the former discussion, RuPt and RuPt/TiO2 are two different catalytic species, the latter featuring a rather more pronounced primary oxide spillover effect. Since hypo-d-oxides, primarily anatase titania, zirconia, and hafnia, facilitate spillover of M-OH, Figure 9 clearly shows the overall effect and advantages of a membrane type OH- - transferring within TiO2 - network of catalyst support (SMSI), relative to the plain carbon. In other words, while Ru itself facilitates Pt-OH and

Neophytides et al. Ru-OH spillover transfer in RuPt composite electrocatalysts,73 the supporting effect of titania advances the whole same effect for more than 200 mV. This remarkable result is one of the most significant confirmations of the present interactive catalytic model, as implemented in electrocatalysis. There are two ways to improve CO tolerance: (a) enable catalytic CO hydrogenation or (b) shift the reversible primary oxide growth closer to and ideally within the potential range of H-adatom anodic oxidation, enabling in this way parallel anodic CO oxidation (eq 10). In such a respect, Magneli phases, and in particular their titania modified phases, impose both a much advanced conductivity and a substantially reduced IR drop, thereby almost theoretical current collection of all properly deposited (d-d-bonded) metal nanostructured electrocatalytic particles. In this way, Magneli phase based supported electrocatalysts approach and even exceed heterogeneous catalysts in their efficiency. 3.5. SMSI and Grafting of Intermetallic Hypo-Hyper-dInterelectronic Type Electrocatalysts. Since the VIA period of elements (Cr, Mo, W) features less pronounced oxophilic properties than the IVA group (Ti, Zr, Hf),32 the former affords much milder experimental conditions to prove the present hypohyper-d-interelectronic and interionic bonding theory and stoichiometry. In such a respect, while MoO3 reduction in a hydrogen stream requires about 730 °C (Figure 10a) and NiO reduces below 260 °C (Figure 10b), their intermetallic phases mostly appear between 300 and 400 °C for all combinations along their phase diagram, whereas reduction of the last amounts of nonstoichiometric bonded MoO3 never approaches individual peak of the latter (Figure 10c). Co, as a metal, which involves one d-electron less than Ni, in such a hypo-hyper-d-interionic interaction, produces the same type of TPR but is shifted for about 100 °C toward higher temperatures, which agrees perfectly well with basic statements of the extended Brewer theory.1-4,8-11,16,25 The differences are even less pronounced in combinations of Ni and Co with W, since the oxide of the latter (WO3) reduces at about 600 °C but follows the Brewer theoretical intermetallic bonding prediction and stoichiometry. A rather fascinating piece of experimental evidence is that, due to the hypo-hyper-d-interionic interaction, these homogeneously mixed oxides thermally interact to produce their stoichiometrically correct and well-defined intermetallic phases far below their individual melting points, and this is the decisive point for all low temperature annealing in supported catalyst production. Such hypo-hyper-d-interionic interaction has been the basis for selective interactive (SMSI) grafting (anchoring) of individual or composite, prevailingly hyper-d-interelectronic nanostructured electrocatalysts upon proper hypo-d-oxide supports. The average nanosize is then a priori determined by the ratio of available exposed selective 5-fold coordinated Ti atom positions in the anatase network or Magneli phases, and available content of interactive mixture of species resulting in final clusters. The other approach consists of composite metal catalysts grafting from the homogeneous colloidal distribution of nanostructured precursors in buffered alkaline (Na2CO3) ethylene glycol (EG) to obey the requirement of acidic (pH 4-6) pzc of Magneli phases and anatase titania interactive supports, or similar polar solvents for negatively charged colloidal nanoprecursors.56,66 The whole hypo-hyper-d-interelectronic theory of electrocatalysis relies on the fact that the stronger the hypo-hyperd-intermetallic bonding, the weaker the intermediate Hadatom adsorption, and thereby the higher the electrocatalytic


J. Phys. Chem. B, Vol. 110, No. 7, 2006 3041 proven at the nanosize scale (Figure 3), while there are still some kinetic constraints to producing quantitatively Brewer type Laves intermetallic phases at the nanostructured scale. Metallurgically properly prepared HfPd3, ZrPt3, and HfPt3 intermetallics belong both to the most stable Laves phases and the most electrocatalytically active specimens for the HER.74 The discrepancies are that it is a particular hard task to produce the same intermetallic phases nanostructured upon hypo-d-oxide supports. In other words, the spillover effect of primary oxides (M-OH) that dynamically advances the electrocatalytic reactions becomes a detrimental obstacle when one has to reduce hypo-d-oxide components according to the above thermodynamic considerations and stoichiometry. Such an approach needs proper interactive catalysts and crystallization ingredients. The resulting consequences are that whereas WPt3 is insoluble even in aqua regia,75 a rather high oxidation resistance enables a ZrPt3 coating to be employed as a protection lining at a 2800 °C direct oxidizing flame in aircraft jets,76,77 with no oxide appearance before or afterward, as shown by XPS. In the same context, although the Brewer HfPd3 stoichiometric compound is known to be an extraordinarily stable intermetallic phase for which the synthesis proceeds with explosive bonding,1,2 none of these materials can be quantitatively produced in nanostructured sizes upon hypo-d-oxide supports by the above-predicted thermal procedure. The latter needs further catalytic and structural refinements. Finally, it is worth mentioning that Tseung et al.78,79 were the first to produce the same type of Pt/WO3 interactive hypod-oxide supported electrocatalysts with remarkable catalytic effect for both the HER and CO tolerance. Pt was co-deposited from a tungstic acid solution, but the authors ascribed to such a catalyst the properties of a bronze. 4.0. Conclusions

Figure 10. 10. TG grams of TPR of individual and 1:1 molecular level mixture of Mo-Ni oxides in a hydrogen stream to produce Mo and Ni metals and their defined intermetallic phase. Temperature scan 5 °C/min.

activity.8-11,16,25 The same conclusion concerns the SMSI of prevailingly hyper-d-intermetallic phases upon hypo-d-oxide supports, while the present method combines these two into the one overall supported (SMSI) composite hypo-hyper-d-interactive electrocatalytic effect. In other words, the hypo-hyper-dinterionic effect imposes itself not only within the intermetallic phase of catalytic clusters but also in its bonding interaction with the hypo-d-oxide support, too. Thus, Pt/C and Pt/TiO2/C (or Pt/Ti2nO(2n-1)) are two different types of catalytic species. While intermetallic interactions of the VIA group of metals proceed stoichiometrically and produce intermetallic phases of extraordinary stability, the problem with more oxophilic elements is rather more complex. Thermodynamic aspects related to the IVA group of hypo-d-transition metals in their interaction with hyper-d-electronic elements are XRD, XPS and analytically

Magneli phases are shown to have unique SMSI properties, rather high electronic conductivity, extraordinary chemical stability, and interactive spillover features for primary oxides, thus enabling composite synergistic electrocatalysts of outstanding activity and durability to be created. The weak point of market supplied Magneli phases is their low specific surface area per unit mass. Grinding in planetary mills enables it to be increased by 1 order of magnitude (∼1-3 m2 g-1) and still keep the same properties. A submonolayer of anatase titania enables the available SMSI surface to be increased for 1 more order of magnitude and still keep the same conductive but advanced interactive catalytic properties. The present authors expect to be able to produce nanostructured (50 nm) Magneli phases by direct reduction of nanosized colloidal titania precursors, instead of the so far used catalytic (Pd) process of production. Such material is far superior as compared with carbon supports of catalysts, since the latter, for example, can by no means be employed as anodic carriers in water electrolysis (hydrogen generation in membrane cells), because of its rather facilitated oxidation by evolving oxygen. The more developed sol-gel produced carbon particles, the higher is their sensitivity to oxidation. Carbon is a noninteractive conductive support of higher dispersion, Magneli phases are better (electronic) conductors, d-d-bonding and current collectors for every single nanostructured metallic particulate catalyst, highly interactive and extremely stable materials. LT PEM FCs could not be further imagined in any advanced form without Magneli phases as highly electronically conductive and perfect current collecting nanomaterials, SMSI in d-d-bonding of composite electrocatalysts, and thereby providing their synergistic activity, otherwise

3042 J. Phys. Chem. B, Vol. 110, No. 7, 2006 advanced by the interactive primary oxide (M-OH) spillover and participation in electrocatalytic reactions. Acknowledgment. This paper is dedicated to the memory of distinguished Professor Leo Brewer, Department of Chemistry, University of California at Berkeley, who recently passed away, and whose Intermetallic bonding theory has been for more than two decades in the core and basis of the electrocatalytic model employed by one of the authors (M.M.J.). This paper has been supported by and carried out within EU Project ‘Apollon’, contract Nr.ENK5-CT-2001-00572, EU Project NR. NNES-2001-00187, and the Project ‘Prometheas’, contract Nr.ICA2-2001-10037. References and Notes (1) Brewer, L. Science 1968, 161, 115. (2) Brewer, L. In Electronic Structure and Alloy Chemistry of Transition Elements; Beck, P. A., Ed.; Interscience: New York, 1963; p 221. (3) Brewer, L. In High-Strength Materials; Zackay, V. F., Ed.; Wiley: New York, 1965; p 12. Brewer, L. The CohesiVe Energies of the Elements; LBL-3720; Berkeley, California, May 4, 1977. (4) Brewer, L. In Phase Stability in Metals and Alloys; Rudman, P., Stringer, J., Haffee, R. I., Eds.; McGraw-Hill: New York, 1967; p 39. (5) Friedel, J.; Sayers, C. M. J. Phys. 1977, 38, 697. (6) Friedel, J. In Physics in Metals, I: Electrons; Ziman, J. M., Ed.; Cambridge University Press: Cambridge, U.K., 1969; p 340. (7) De Boer, F. R.; Boom, R.; Mattens, W. C. M.; Miedema, A. R.; Niessen, A. K. In Cohesion and Structure, Cohesion in Metals, Transition Metal Alloys; De Boer, F. R., Pettifor, D. G., Eds.; North-Holland: Amsterdam, The Netherlands, 1988; Vol 1. (8) Jaksic, M. M. Electrochim. Acta 1984, 29, 1539. (9) Jaksic, M. M. J. Mol. Catal. 1986, 38, 161. (10) Jaksic, M. M. J. New Mater. Electrochem. Syst. 2000, 3, 153. (11) Jaksic, M. M. Electrochim. Acta 2000, 45, 4085. (12) Kita, H. J. Elctrochem. Soc. 1966, 113, 1095. (13) Miles, M. H. J. Electroanal. Chem. 1975, 60, 89. (14) Trasatti, S. In AdVances in Electrochemistry and Electrochemical Engineering; Tobias, C. W., Gerischer, H., Eds.; Interscience: New York, 1977; Vol 10, p 213. (15) Pauling, L. The Nature of Chemical Bond, 3rd ed.; Cornel University Press: Ithaca, NY, 1960. (16) Neophytides, S. G.; Zafeiratos, S.; Papakonstantinou, G. D.; Jaksic, J. M.; Paloukis, F. E.; Jaksic, M. M. Int. J. Hydrogen Energy 2005, 30, 131, 393. (17) Trasatti, S. J. Electroanal. Chem. 1971, 33, 351. (18) Trasatti, S. Z. Phys. Chem. N. F. 1975, 98, 75. (19) Methfessel, M.; Hennig, D.; Scheffler, M. Phys. ReV. B 1992, 46, 4816. (20) Gschneidner, K. A. In Solid State Physics; Seitz, F., Turnbull, D., Eds.; Acadenic Press: New York, 1964; Vol. 16, p 275. (21) Allen, B. C. Trans. Met. Soc. AIME 1963, 227, 1175. (22) Sabatier, P. La Catalyse en Chimie Organique; Librairie Polytechnique: Paris, 1913. Sabatier, P. Ber. Dtsch. Chem. Soc. 1911, 44, 2001. (23) Fukashima, H.; Akiyama, T.; Akagi, S.; Higashi, K. Trans. Jpn. Inst. Met. 1979, 20, 358. (24) Younes-Metzler, O.; Zhu, L.; Gileadi, E. Electrochim. Acta 2003, 48, 2551. (25) Neophytides, S. G.; Zafeiratos, S., Jaksic, M. M. J. Electrochem. Soc. 2003, 150, E512. (26) Zafeiratos, S., Palukis, F. E.; Jaksic, M. M.; Neophytides, S. G. Surf. Sci. 2004, 552, 215. (27) Rodriguez, J. A.; Goodman, D. W. Science 1992, 257, 897. (28) Kubaschewski, O.; Alcock, C. B.; Spenser, P. J. Materials Thermochemistry, 6th ed.; Pergamon Press: Oxford, U.K., 1993. (29) Meschter, P. J.; Worrell, W. L. Metall. Mater. Trans. A 1977, 7, 299. (30) Meschter, P. J.; Worrell, W. L. Metall. Mater. Trans. A 1977, 8, 503. (31) Magneli, A. Acta Chem. Scand. 1959, 13, 5. (32) Ley, K. L.; Liu, R.; Pu, C.; Fan, Q.; Leyarovska, N.; Segre, C.; Smotkin, E. S. J. Electrochem. Soc. 1997, 144, 1543. (33) Zafeiratos, S.; Papakonstantinou, G.; Jaksic, M. M.; Neophytides, S. G. J. Catal. 2005, 232, 127. (34) Tauster, S. J.; Fung, S. C. J. Catal. 1978, 55, 29.

Neophytides et al. (35) Tauster, S. J.; Fung, S. C.; Baker, R. T. K.; Horsley, J. A. Science 1981, 217, 1121. (36) Tauster, S. J.; Fung, S. C.; Garten, R. L. J. Am. Chem. Soc. 1978, 100, 170. (37) Stevenson, S. A. Metal-Support Interactions in Catalysis, Sintering and Redispersion; Van Nostrand: New York, 1987. (38) Hegedus, L. L., Ed. Catalysis Design; Wiley: New York, 1987. (39) Haller, G. L.; Resasco, D. E. AdV. Catal. 1989, 36, 173. (40) Nicole, J.; Tsiplakidis, D.; Pliangos, C.; Verykios, X. E.; Comninellis, C.; Vayenas, C. G. J. Catal. 2001, 204, 23. (41) Vayenas, C. G.; Bebelis, S.; Pliangos, C.; Brosda, S.; Tsiplakidis, D. Electrochemical ActiVation of Catalysts, Promotion, Electrochemical Promotion, and Metal-Support Interaction; Kluwer: New York, 2001. (42) Haruta, M. Catal. Today 1997, 36, 153. (43) Date, M.; Haruta, M. J. Catal. 2001, 201, 221. (44) Boccuzzi, F.; Chiorino, A.; Tsubota, S.; Haruta, M. J. Catal. 2001, 202, 256. (45) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Lu, P.; Akita, T.; Ichikawa, S.; Haruta, M. J. Catal. 2001, 202, 256. (46) Kung, H. H.; Kung, M. C.; Costello, C. K. J. Catal. 2003, 216, 425. (47) Akita, T.; Lu, P.; Ichikawa, S.; Tanaka, K.; Haruta, M. Surf. Interface Anal. 2001, 31, 73. (48) Haruta, M. Chem. Rec. 2003, 3, 75. (49) Haruta, M. Catal. SurV. Jpn. 1997, 1, 61. (50) Greeley, G.; Mavrikakis, M. Nat. Mater. 2004, 3, 810. (51) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647. (52) Mavrikakis, M.; Stoltze, P.; Norskov, J. K. Catal. Lett. 2000, 64, 101. (53) Christoffersen, E.; Liu, P.; Ruban, A.; Skriver, H. L.; Norskov, J. K. J. Catal. 2001, 199, 123. (54) Hammar, B.; Norskov, J. K. AdV. Catal. 2000, 45, 71. (55) Zhou, W.; Zhou, Z.; Song, S.; Li, W.; Sun, G.; Tsiakaras, P.; Xin, Q. Appl. Catal., B 2003, 46, 273. (56) Zhou, W. J.; Li, W. Z.; Song, S. Q.; Zhou, Z. H.; Jiang, L. H.; Sun, G. Q.; Xin, Q.; Poulianitis, K.; Kontou, S.; Tsiakaras, P. J. Power Sources 2004, 131, 217. (57) Kudo, T. Nature (London) 1984, 312, 537. (58) Livage, J.; Henry, M.; Sanchez, C. Prog. Solid State Chem. 1988, 18, 259. (59) Opra-Krasovac, U.; Surca-Vuk, A.; Orel, B. Electrochim. Acta 2001, 46, 1921. (60) Bock, C.; Paquet, C.; Couillard, M.; Botton, G. A.; MacDougall, B. R. J. Am. Chem. Soc. 2004, 126, 8028. (61) Dauscher, A.; Hilaire, L.; Spirlet, J. C.; Muller, W.; Maire, G. Surf. Sci. 1988, 204, 161. (62) Cameron, S. D.; Dwyer, D. J. J. Vac. Sci. Technol. 1987, A5, 651, 57. (63) Vracar, Lj. M.; Krstajic, N. V.; Radmilovic, V. R.; Jaksic, M. M. J. Electroanal. Chem. 2005, 587, 99. (64) Aita, T.; Tanaka, K., Tsubota, S.; Haruta, M. J. Electron Microsc. 2000, 49, 657. (65) Arabatzis, I. M.; Stergiopoulos, T.; Andreeva, D.; Kitova, S.; Neophytides, S. G.; Falaras, P. J. Catal. 2003, 220, 127. (66) Arabatzis, I. M.; Antonaraki, S.; Stergiopoulos, T.; Hiskia, A.; Papaconstantinou, E.; Bernard, M. C.; Falaras, P. J. Photochem. Photobiol., A 2002, 149, 237. (67) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (68) Jaksic, M. M. Science, to be submitted for publication, 2005. (69) Vittadini, A.; Selloni, A.; Rotzinger, F. P.; Gratzel, M. Phys. ReV. Lett. 1998, 81, 2954. (70) Lazzeri, M.; Vittadini, A.; Selloni, A. Phys. ReV. B 2001, 63, 155409. (71) Tilocca, A.; Selloni, A. J. Chem. Phys. 2003, 119, 7445. (72) Li, W.; Liang, C.; Zhou, W.; Qiu, J.; Zhou, Z.; Sun, G.; Xin, Q. J. Phys. Chem. B 2003, 107, 6292. (73) Watanabe, M.; Motoo, S. J. Electroanal. Chem. 1975, 60, 267. (74) Jaksic, M. M. Mater. Chem. Phys. 1989, 22, 1. (75) Toshiaki, T.; Tsutomu, Y.; Shigeo, K.; Kenjiro, Y.; Matunori, S.; Tanaka Kikinzoku Kogyo Industry, Tokyo, Japan, Jap. Pat. 4-32595, May 29, 1990. (76) Alvey, M. D.; George, P. M. Carbon 1991, 29, 523. (77) Fisher, R. F.; Alvey, M. D.; George, P. M. J. Vac. Sci. Technol. 1992, 10, 2253. (78) Tseung, A. C. C.; Shen, P. K.; Chen, K. Y. J. Power Sources 1996, 61, 223. (79) Tseung, A. C. C.; Chen, K. Y. Catal. Today 1997, 38, 439.

Composite Hypo-Hyper-d-Intermetallic and Interionic Phases as

Recommend Documents