Dynamic Fluctuation of Activity in Fe/N/C Oxygen Reduction Reaction Catalyst Depending on Heat Treatment Time

Abstract

Introduction Fe/N/C catalyst is one of the most studied non-precious metal catalysts for oxygen reduction reaction (ORR). It usually prepared by heat treatment of precursor, such as Fe-N containing polymers or chelates, with or without carbon support. The relation of heat treatment time and ORR activity in Fe/N/C catalyst has been studied from some aspects of view. For example, Kamiya et al. reported existence of an optimum heat treatment time, 45s, investigating catalyst prepared by heat treatment of Fe–pentaethylenehexamine complex and graphene oxide under Ar [1]. It is proposed that Fe-N active sites are initially formed on graphene oxide and then decomposed during heat treatment resulting in the existence of an optimum heat treatment time. In this study, investigating catalyst prepared by heat treatment of hemin, a Fe-porphyrin derivative, and carbon black support under Ar, we unveil a dynamic fluctuation of activity and Fe-N active site content with more than one local maximum with the increase in heat treatment time. These behaviors can be explained by a reaction model considering multiple types of Fe-N active site with individual formation and decomposition rate constant. It could be a support for existence of multiple types of Fe-N active site responsible for ORR activity and gives us a comprehensive understanding how net ORR activity is achieved by constituent active site formed and decomposed during heat treatment. Experimental For preparing catalyst, 200 mg of hemin was mixed with 200 mg of carbon black (Ketchen Black 600JD, Lion Corp.) in dimethylformamide. After ultrasonicating the solution for 30 mins, the dried mixture was obtained by a rotary evaporator and heat treated under Ar by different patterns of heat treatment time: 45s, 100s, 5m, 30m, 2hrs, 5hrs or 10hrs at temperature of 750°C. Finally, the mixture was acid washed in 2M H2SO4(aq). Catalyst's ORR activity was evaluated in 0.5M H2SO4 (aq) by rotating disk electrode (RDE) experiments. Catalyst was coated on a glassy carbon working electrode with the area of 0.2826 cm2 to be 0.57mg/cm2 catalyst loading. Pt and reversible hydrogen electrode (RHE) generated using the same electrolyte were used as the counter and reference electrode respectively. The rotating speed of the working electrode was set to 1500 rpm. Kinetic current density, ik, is compared, where ik = iL*i/(iL -i). Here, i is observed current density at 0.8V and iL is diffusion limit current density defined as current density at 0.4 V vs. RHE. X-ray photoelectron spectroscopy (XPS, PHI 5000 Versaprobe) measurements were performed. Fe 2p3/2 spectrum is observed for estimation of Fe-N active site content in catalyst, and C1s spectrum was collected to calibrate Fe 2p3/2 peak intensity. Results and Discussion Figure.1a shows the Fe 2p spectrum and its fitting result for the sample with heat treatment time of 5 hrs.The spectrum was well fitted and separated into two peaks originated from Fe in hemin-like structure and Fe-N active site formed on carbon support. Peak position and FWHM of Fe in hemin-like structure are set to 711.25 eV and 5.1 eV, respectively. Fe-N active site peak position and FWHM were set free to move through peak separation fitting to collectively represent several possible types of Fe-N active site structures, FeN2, FeN4 or FeN3, as one peak. Its peak position and FWHM exist in the range of 709.0-710.2eV and 2.15-3.06 eV, respectively. Fig. 1b shows the behavior of activity and content of Fe-N active site normalized by that of C. We can see more than one local maximum in activity with the increase in heat treatment time. One local maximum is located around 5m and another exists around 5hrs. Only by investigating over such a wide range of heat treatment time, we can realize existence of multiple local maximums. The discrepancy for the position of the first local maximum with that of Kamiya et al. [1] may come from difference in choice of precursor. Fe-N active site content shows similar behavior to that of activity, which can be fitted by two curves with the expression of A[exp(-kf*t)-exp(-kd*t)]. We attribute them to formation and decomposition of two different types of Fe-N active site, (Fe-N)α and (Fe-N)β with individual formation rate, kf, and decomposition rate, kd. It could be a support for existence of more than one type of Fe-N active site responsible for ORR activity. Detailed reaction model and structure of two active sites will be discussed in session.