Journal of Electroanalytical Chemistry Journal of Electroanalytical Chemistry 576 (2005) 57–63 www.elsevier.com/locate/jelechem A reaction route graph analysis of the electrochemical hydrogen oxidation and evolution reactions I. Fishtik a a,* , C.A. Callaghan a, J.D. Fehribach b, R. Datta a,* Department of Chemical Engineering, Fuel Cell Center, Worcester Polytechnic Institute, 100 Institute Road, Worcester, MA 01609, USA b Department of Mathematical Sciences, Worcester Polytechnic Institute, Worcester, MA 01609, USA Received 30 June 2004; received in revised form 24 August 2004; accepted 13 October 2004 Abstract The reaction route (RR) graph approach recently developed by us for complex, non-linear kinetic mechanisms is applied to the hydrogen oxidation and evolution reactions. A corresponding RR graph is constructed and translated into an equivalent electrical circuit network by associating each elementary step with a characteristic resistance for the steady-state case and considering the overall reaction as a power source. It is further shown that the steady-state kinetics of the reaction can be investigated employing the conventional methods of the electrical network theory. Using a set of rate constants for the hydrogen evolution reaction (her) in alkaline solutions from the literature, the dominant RRs are identified and simplified mechanisms and kinetics derived. Ó 2004 Elsevier B.V. All rights reserved. Keywords: Hydrogen oxidation reaction; Hydrogen evolution reaction; Graph theory; Electrical networks 1. Introduction Graph theoretical methods are being increasingly utilized to study the topology of chemical reaction networks [1]. In electrochemical reaction systems, an appropriate topological description of the mechanism is especially important since it is directly related to the corresponding electrical circuit that is used to analyze and interpret impedance spectroscopy data [2]. It is, therefore, of interest to have a methodology for translating a reaction mechanism into an equivalent circuit. Graph theoretical methods are of interest in this regard. Normally, a chemical or electrochemical reaction mechanism is depicted graphically by placing individual species at the nodes of a graph while the connectivity of the nodes reflects the stoichiometry of the mechanism * Corresponding authors. Tel.: +50 883 15445; fax: +50 883 15853. E-mail addresses: ifishtik@wpi.edu (I. Fishtik), rdatta@wpi.edu (R. Datta). 0022-0728/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jelechem.2004.10.007 [1]. Although such an approach is useful in studying many topological aspects of certain reaction mechanisms, it presents difficulties for the more general case and leaves open several significant kinetic issues such as reduction and simplification. We have recently proposed [3–6] a new, more general, graph theoretical approach applicable to complex, nonlinear kinetic mechanisms. The main idea is to represent a kinetic mechanism by a reaction route (RR) graph such that the branches are directly associated with the elementary steps while the nodes represent simply their connectivity and also satisfy the steady-state conditions for the reaction intermediates and terminal species. As a result, we obtain a graph that is particularly useful and has a remarkable property, namely, any walk between two terminal nodes in this graph is a full RR (FR), i.e., an RR that produces an overall reaction (OR). For the steady-state case, we further employed the concept of the resistance (impedance for the more general case) of a chemical reaction defined as the ratio between 58 I. Fishtik et al. / Journal of Electroanalytical Chemistry 576 (2005) 57–63 its dimensionless affinity and rate and, thus, showed that there is a complete analogy between RR graphs and electrical networks. Altogether, this is a powerful new approach for comprehension, analysis, reduction and simplification of complex kinetic mechanisms, especially in relation to impedance spectroscopy. In this contribution, we apply the RR graph approach to the steadystate analysis of the electrochemical hydrogen oxidation (hor) and evolution (her) reactions. ! 2 rT ¼ k T h2H k T ð1 hH Þ ; ð8Þ ! b FE ð1 bV ÞFE rV ¼ k V ð1 hH Þ exp V k V hH exp ; RT RT ð9Þ ! bH FE ð1 bH ÞFE k H ð1 hH Þ exp : rH ¼ k H hH exp RT RT ð10Þ 2. The mechanism and elementary reactions The hor and her reactions are by far the most thoroughly investigated electrochemical reaction systems [7,8] and have assumed a new importance by virtue of renewed interest in fuel cells. The mechanisms of these processes are fairly well established and, as is widely accepted, are described by three elementary reactions referred to as the Volmer (sV), Heyrovsky (sH) and Tafel (sT) steps. Thus, in acidic solutions the mechanism of the hor is sT: H2 þ 2M¡2HM ð1Þ sV: H2 O þ HM¡M þ H3 Oþ þ e ð2Þ sH: H2 O þ H2 þ M¡HM þ H3 Oþ þ e ð3Þ where sq is a shorthand notation for an elementary step q and M is an active site on the electrode surface. The OR for the hor in acidic solutions is OR: H2 þ 2H2 O $ 2H3 Oþ þ 2e ð4Þ The elementary steps in the her are the reverse of those describing the hor. For instance, in alkaline solutions, the mechanism of the her is sT: 2HM¡2M þ H2 ð5Þ sV: M þ H2 O þ e ¡HM þ OH ð6Þ sH: HM þ H2 O þ e ¡M þ H2 þ OH Here, hH is the hydrogen surface coverage, ! ! k V ; k V ; k H ; k H ; k T and k T are the chemical pseudo-rate constants (in mol cm2 s1) as defined in [9], E is the potential referred to the standard hydrogen electrode scale (SHE) and the other notation is conventional. The composition dependence of fluid species is included in the pseudo-rate constants. Thus, ! ð7Þ Hence, for the her in alkaline solutions the OR is OR: 2H2 O þ 2e ¡H2 þ 2OH Thus, while H3O+ is the proton source for the her in acidic media, H2O is the proton source in alkaline media. In neutral solutions, both acidic and basic hydrogen reactions can occur. Since the mechanisms of both the her and the hor are similar, in what follows, for simplicity, we will consider only the her in the alkaline solutions, i.e., elementary steps (5)–(7). The net rates of these steps, i.e., the rates ! of forward minus reverse steps, rq ¼ r q r q , on nickel, in alkaline solutions are described by the following equations [9]: !0 ! 0 k V ¼ k V c H2 O ; k V ¼ k V cOH ; ! !0 k H ¼ k H c H2 O ; 0 k H ¼ k H cOH ðP H2 =P 0 Þ !0 0 where k q and k q are the partial standard reaction rate constants (in cm s1). The elementary steps are characterized by their (dimensionless) affinities defined as [10], ! 1 rq Aq ¼ ln : Aq ¼ RT rq ð11Þ Thus, ! AT ¼ 1 k T h2H AT ¼ ln ; 2 RT k T ð1 hH Þ ð12Þ ! V FE k V ð1 hH Þ exp bRT 1 AV ¼ ln AV ¼ h i ; RT k h exp ð1bV ÞFE ð13Þ ! k H hH exp bHRTFE 1 AH ¼ ln AH ¼ h i: RT k H ð1 hH Þ exp ð1bRTH ÞFE ð14Þ V H RT We also define the resistances of the individual elementary steps as [3,11], ! Rq Aq lnð r q = r q Þ ¼ ! ; rq rq rq ð15Þ i.e., ! ln k T h2H AT k T ð1hH Þ2 RT ¼! ; rT 2 k T hH k T ð1 hH Þ2 ð16Þ I. Fishtik et al. / Journal of Electroanalytical Chemistry 576 (2005) 57–63 RV AV rV ! ln ð1b k V hH exp while FRVH and FRVT are familiar, FRHT is less so. These relations imply a similar set of equations between the affinity AOR of the OR and the affinities AV ; AH and AT of the elementary steps, i.e., bV FE RT V ÞFE RT k V ð1hH Þ exp ¼! h i; V FE k V ð1 hH Þ exp bRT k V hH exp ð1bRTV ÞFE AV AH þ AT ¼ 0; ð17Þ RH AH rH ! ln b FE H RT ð1b ÞFE k H hH exp k H ð1hH Þ exp H RT ¼! h i: k H hH exp bHRTFE k H ð1 hH Þ exp ð1bRTH ÞFE ð18Þ It may be noticed that the resistances are necessarily positive quantities, and are the same for both the her and the hor, i.e., they are independent of the direction of the net reaction. The direction depends on the sign of the affinity, determined largely by the potential E along with species concentrations. Since the affinities are defined as dimensionless quantities, the units of the resistances are rate1. 3. Reaction routes and KirchhoffÕs laws At steady-state, the rate rOR of the OR, expressed as the rate of H2 production for the her, is rOR ¼ rH þ rT ; ð19Þ while the steady-state rate of production and consumption of the surface intermediate HM is equal to zero, i.e., rV rH 2rT ¼ 0: 59 ð20Þ Solving Eq. (20) for rH and rT and substituting the results into Eq. (19) we arrive at two alternate expressions for the rate of the OR: rOR ¼ rV rT ; ð21Þ 2rOR ¼ rV þ rH : ð22Þ Eqs. (19)–(22) are the equivalent of KirchhoffÕs current law (KCL) [3]. As shown by Milner [12] and Happel and Sellers [13], the her involves one cycle, or empty route (ER), and three FRs. These are: ER: sV sH þ sT ¼ 0; ð23Þ FRVH: sV þ sH ¼ OR; ð24Þ FRVT: 2sV þ sT ¼ OR; ð25Þ FRHT: 2sH sT ¼ OR; ð26Þ ð27Þ and AOR ¼ AV þ AH ¼ 2AV þ AT ¼ 2AH AT ; ð28Þ which are the equivalent of KirchhoffÕs voltage law (KVL) [3]. An appropriate RR graph for the system must be consistent with KCL and KVL relations given above. Further, the affinity Aq and reaction rate rq for a given reaction step sq must remain the same, even if the step occurs more than once in an RR graph, as in the case for non-minimal RR graphs. 4. The RR graph By definition [3,5], an RR graph involves two types of nodes. One of these, referred to as terminal nodes (TNs), satisfies the steady-state condition for the OR. In other words, at TNs, the algebraic sum of the rates of the elementary steps leaving or entering the node should equal the rate of the OR. In our case, these conditions are alternatively expressed by Eqs. (19), (21) and (22). Thus, the possible connectivity of the reactions at the TNs is TN1: OR sH sT ð29Þ TN2: OR sV þ sT ð30Þ TN3: 2OR sV sH ð31Þ The other type of nodes, referred to as intermediate nodes (INs), must satisfy the steady-state conditions for the intermediates. That is, the algebraic sum of the rates of the elementary steps leaving or entering an IN should be equal to zero. In our case, this condition is described by Eq. (20). Since for the case of the her, there is only one linearly independent intermediate (HM), we have only one independent IN, and the connectivity of the reactions at this IN is IN : sV þ sH þ 2sT ð32Þ It should be noticed that, although the nodes are defined in terms of the rates of the reactions entering and leaving the nodes at quasi-steady state, i.e, KCL, they are represented in terms of connectivity of reactions sq and OR, i.e., their number and direction. Based on Eqs. (23)–(32), the RR graph may be obtained following the algorithm for RR graph construction described previously [3,5]. Namely, in the particular case of reaction mechanisms in which the number of elementary reactions and ORs in every FR, ER, IN and TN does not exceed ±2, we noticed that the RR graphs are symmetrical and involve each 60 I. Fishtik et al. / Journal of Electroanalytical Chemistry 576 (2005) 57–63 elementary reaction and OR twice [5]. As a result, the realization of the RR graphs may be performed based on the following considerations. The starting point is a set of linearly independent RRs comprising a single minimal smallest FR, the balance being the smallest ERs. The FR is drawn next, along with another variation in which the order of the reactions is reversed. Further, the ERs are placed starting with the smallest ones such that none of the reactions is involved more than once in each of the sub-graphs. The sub-graphs are next fused only via the reactions that are not present in the subgraphs. If all of the smallest ERs have been placed and all nodes, except two, are balanced, the graph is complete. It remains only to balance the remaining two nodes by attaching the OR according to an appropriate TN from the complete list of TNs. Finally, the resulting RR graph should be checked to ensure that all enumerated direct FR and ERs are depicted as walks. The number of linearly independent RRs for the her is equal to 2. Thus, according to the algorithm, these are selected as FRVH, and the ER. The RR graph construction is depicted graphically in Fig. 1. We begin by depicting twice the FRVH, the second with the steps in reverse order. Next, we add symmetrically ER to each of the graphs. At this stage it is seen that, because in each individual sub-graph the ER cannot be introduced into the graph due to the direction of the existing branches, it is necessary to connect the graphs by fusing two nodes. Finally, we identify the balanced INs and potential TNs. The TNs are then balanced via the addition of the OR yielding the RR graph. It may be seen that the resulting graph is an RR graph in that it satisfies all of the conditions imposed on the RR graph. That is, the graph involves the ER, Eq. (23), all of the three FRs, Eqs. (24)–(26), the INs satisfy Eq. (32), while the TNs satisfy Eq. (31). Notice that every FR in the graph is involved twice. This is a consequence of the fact that the mechanism is non-minimal [5], i.e., the elementary steps are involved more than once in an FR. Nonetheless, the affinity and the rate of a step (e.g., sH or sV) remain unchanged regardless of their placement, as, of course, is required in any non-minimal graph. This imparts a certain symmetry to the RR graph, i.e., the branches may be replaced by resistors for steady-state kinetic analysis or by a combination of resistors and capacitors for the general case, depending upon the objective. 5. Reaction kinetics Next, we employ the electrical circuit analogy to analyze the steady-state kinetics of the her. More specifically, our goal is to determine the dominant FRs and, eventually, if possible, to reduce and simplify the mechanism and kinetics. In doing this, the elementary steps are viewed as resistances while the OR is viewed as a power source [3]. The resulting electrical network is presented in Fig. 2. Thus, we are now in a position to employ the conventional electrical network analysis [14], e.g., KirchhoffÕs laws, to evaluate the currents (rates) and resistances through various FRs. It should be noticed that the representation of the reaction steps simply as resistances means that the resulting equivalent circuit is suitable for steady-state dc analysis rather than for an ac analysis. The electrical network equivalent of the RR graph may also be used to interpret and analyze integral impedance spectroscopy measurements, provided that + O R - RH RV RT RT RV RH + Fig. 1. The realization RR graph for her and hor. O R - Fig. 2. The electrical circuit analog of the RR graph. I. Fishtik et al. / Journal of Electroanalytical Chemistry 576 (2005) 57–63 the reactions are represented as an appropriate combination of resistors and capacitors. First, let us show that the electrical network adequately describes the mechanism. Consider the overall resistance ROR of the network. Employing the standard procedures, e.g., a D–Y conversion [14] (Fig. 3), the overall network resistance is readily shown to be equal to: ROR ¼ RA þ ðRV þ RB ÞðRH þ RC Þ ; RV þ RH þ RB þ RC Numerical evaluations of the polarization curve according to Eqs. (38) and (39) are presented in Fig. 4. Here, and in the discussion below, the numerical calculations were performed using a set of chemical rate constants (mol cm2 s1) and partial standard chemical rate constants (cm s1) for individual steps on Ni in 1.0 mol dm3 NaOH at 20.0 °C [8]. Under these conditions the constants are [9]: ð33Þ ! k V ¼ 7:8 1017 ; where k H ¼ 3:2 1019 ; ð34Þ RB ¼ RH ðRT =2Þ ; RV þ RH þ ðRT =2Þ ð35Þ RC ¼ RV ðRT =2Þ : RV þ RH þ ðRT =2Þ ð36Þ AOR AV þ AH : ¼ 2ROR 2 RA þ ðRV þRB ÞðRH þRC Þ RV þRH þRB þRC k T ¼ 1:3 109 F ðAV þ AH Þ B ÞðRH þRC Þ RA þ ðRRVV þR þRH þRB þRC ! kV kT j ¼ F ðrV þ rH Þ: The proof that Eq. (38) is equivalent to Eq. (39) is elementary but cumbersome and, for brevity, is not presented here. In particular, the equivalence of Eqs. (38) and (39) may be easily shown by substituting Eqs. (16)–(18) into Eqs. (33)–(36) and further into Eq. (38), additionally employing Eqs. (20) and (27). RT - + O R O R - + O R - RC RV RV RA RT/2 RH + ð40Þ : kH - RH RT RV kH According to the electrical circuit theory, the dominant FRs under a given set of conditions are those that have the lowest resistances. To determine which of the three possible FRs are dominant, we have performed numerical calculations of the resistances of the elementary steps, as a function of potential E, employing the experimental conditions and rate constants given above. The results are presented in Fig. 5. As can be seen, the Volmer step has the smallest resistance over the entire range of potential. On the other hand, the Tafel step has the highest resistance in the potential range E < 1.20 V while the resistance of the Heyrovsky step is the highest in the potential range E > 1.20 V. It may be further noted that RT is also a function of potential, even though Eq. (16) may lead one to believe otherwise. The reason is the variation of hH with potential, as obtained by solving Eq. (20). ð39Þ RV ¼ 6. Dominant FRs ð38Þ RH ! ð37Þ : O R ! kV kT Now, in order to prove that the electrical network quantitatively describes the kinetics of the her it is necessary to show that Eq. (38) is identical to the conventional rate, i.e., a quasi-state current, + k T ¼ 7:8 1012 : and As can be seen from Fig. 4, there is an exact match between the current density evaluated from the electrical network and the current density evaluated via the conventional way. KVL also provides thermodynamic consistency of the rate constants involved in an ER [9], Notice that the rate has been divided by 2 since the network involves two ORs. The corresponding current density is j ¼ 2FrOR ¼ k H ¼ 3:7 106 ; ! According to OhmÕs law, the rate rOR of the OR may be expressed via the overall affinity AOR and overall resistance ROR of the network via rOR ¼ k V ¼ 1:5 101 ; ! RV RH ; RA ¼ RV þ RH þ ðRT =2Þ 61 RV RH + O R - RB RH + O R Fig. 3. The evaluation of the overall resistance of the electrical network. - 62 I. Fishtik et al. / Journal of Electroanalytical Chemistry 576 (2005) 57–63 Fig. 4. Polarization curves evaluated via different equations (see the text for details). RVH ¼ RV þ RH ; ð42Þ RVT ¼ 4RV þ RT ; ð43Þ RHT ¼ 4RH þ RT : ð44Þ Numerical simulations of these quantities, presented in Fig. 6, clearly show that the FRVH dominates the kinetics at low potentials while FRVT is dominant at higher potentials. Further, RVH RH and RVT RT, i.e., these steps dominate the FR total resistances. Based on the numerical evaluations, the electrical network may be reduced and simplified as shown in Fig. 7. More specifically, for potentials E > 1.20 V the kinetics are dominated by the RVT, i.e., sV : M þ H2 O þ e ¡HM þ OH 2 sT : 2HM¡2M þ H2 1 ————————————————— OR : 2H2 O þ 2e ¡H2 þ 2OH where the numbers on the right are stoichiometric numbers. The current density corresponding to this FR may be accurately described by j ¼ 2FrOR ¼ Fig. 5. Resistances of the elementary steps vs. potential. According to the principle of extremum resistance [6], thus, it may be expected that at E < 1.20 V the Tafel step may be neglected since its resistance is much higher as compared to the resistances of the Volmer and Heyrovsky steps. Furthermore, under these conditions, the resistance of the Volmer step is lower than the resistance of the Heyrovsky step, which means that the rate of the reaction is controlled by the Heyrovsky step. A different situation is observed for E > 1.20 V. As can be seen from Fig. 5, in this region the Heyrovsky step has the highest resistance and, consequently, the effect of this step can be neglected. As a result, the kinetics of the reaction may be accurately described by the Tafel and Volmer steps, the Tafel step being rate controlling since its resistance is higher. These conclusions are in complete agreement with those of Krstajic et al. [9]. The same conclusions result for the her in alkaline solutions on Ni if the total resistances of the individual FRs are considered [6]. In defining these we take into account the fact that the elementary steps in an individual RR are connected in series and each elementary step sq in the sequence occurs rq times, where rq is the stoichiometric number of a step in an FR. Thus, the resistance RFR of an FR is [6], X RFR ¼ r2q Rq : ð41Þ 2F ðAV þ AT Þ : 2RV þ RT =2 ð45Þ On the contrary, at higher potentials, e.g., E < 1.20 V, the kinetics are dominated by RVH, i.e., sV : M þ H2 O þ e ¡HM þ OH 1 sH : HM þ H2 O þ e ¡M þ H2 þ OH 1 ——————————————————— OR : 2H2 O þ 2e ¡H2 þ 2OH Respectively, under these conditions, the overall current density, Eq. (38), is reduced to j ¼ 2FrOR ¼ 2F ðAV þ AH Þ : RV þ RH The performance of these reduced rate equations is compared in Fig. 4. Thus for E < 1.20 V, Eq. (46) is accurate, while for E > 1.0 V, Eq. (45) is accurate. In the range of 1.20 V < E < 1.0 V the complete rate equation, Eq. (38), is required. q The resistances of the three FRs that describe the her are, thus: ð46Þ Fig. 6. Resistances of the FRs vs. potential. I. Fishtik et al. / Journal of Electroanalytical Chemistry 576 (2005) 57–63 + O R 63 - RH RV RT RT RV RH + O R - RT RV + -1.5 RV RH O R - RT + O R RV - -1.1 -0.9 E/V Fig. 7. Reduction and simplification of the electrical network as a function of potential. 7. Conclusions From the analysis presented above, it follows that the RR graph approach may be successfully applied to electrochemical reaction systems. Thus, the three elementary steps describing the mechanism of the hor and her are represented as an RR graph. The latter can be further translated into an equivalent electrical network. Thereupon, one may employ the well-developed and powerful methods used in conventional electrical networks to study the kinetics of the hor and her mechanisms, e.g., to determine the dominant FRs. References [1] O.N. Temkin, A.V. Zeigarnik, D.G. Bonchev, Chemical Reaction Networks: A Graph-Theoretical Approach, CRC Press, New York, 1996. [2] D.A. Harrington, P. van den Driessche, J. Electroanal. Chem. 567 (2004) 153. [3] I. Fishtik, C.A. Callaghan, R. Datta, J. Phys. Chem. B 108 (2004) 5671. [4] I. Fishtik, C.A. Callaghan, R. Datta, J. Phys. Chem. B 108 (2004) 5683. [5] I. Fishtik, C.A. Callaghan, R. Datta, J. Phys. Chem. B, in press. [6] I. Fishtik, C.A. Callaghan, R. Datta, in: J. Leszczynski (Ed.), Computational Material Science, vol. 15, Elsevier, Amsterdam, 2004. [7] A.K.N. Reddy, J. OÕM. Modern Electrochemistry, Plenum Press, New York, 1973. [8] B.E. Conway, B.V. Tilak, Electrochim. Acta 47 (2002) 3571. [9] N. Krstajic, M. Popovic, B. Grgur, M. Vojnovic, D. Sepa, J. Electroanal. Chem. 512 (2001) 16. [10] T. De Donder, P. Rysselberghe, Thermodynamic Theory of Affinity, Stanford University Press, Palo Alto, CA, 1936. [11] J.S. Shiner, Adv. Thermodyn. 6 (1992) 248. [12] P.C. Milner, J. Electrochem. Soc. 11 (1964) 228. [13] J. Happel, P.H. Sellers, Adv. Catal. 32 (1983) 272. [14] N. Balabanian, T.A. Bickart, Electrical Network Theory, Wiley, New York, 1969.