**1. Introduction**

Interest in renewable energy sources has increased considerably, mainly due to concerns related to the climate change caused by the atmospheric accumulation of greenhouse gases resulting from fossil fuel use [1]. A promising alternative to the fossil fuels used in the transportation sector is the production of biofuels—e.g., biodiesel, green diesel, and biokerosene—from renewable feedstocks, including vegetable oils and animal fats [2,3]. Moreover, to improve the economics of these biofuels and avoid disrupting the food supply, attention has shifted to low-cost inedible feedstocks, including used cooking oil (UCO), which is also known as yellow grease (YG) [4–6]. This particular waste stream is both abundant and inexpensive (~\$463/ton), with ca. one million tons being produced annually in the U.S. alone [7].

Of the biofuels mentioned above, biodiesel is the name given to the fatty acid methyl esters (FAMEs) resulting from the catalytic transesterification of fat, oil, or greases (FOG) with methanol. Biodiesel has several advantages over petroleum-based diesel, such as being biodegradable and producing less harmful gas emissions and particulate matter upon combustion [8,9]; however, it also displays several drawbacks, including poor cold flow properties as well as relatively low thermal and oxidative stability, mostly arising from its oxygen content [10]. The catalytic deoxygenation of FOG to fuel-like hydrocarbons o ffers advantages over biodiesel in terms of fuel quality and feedstock flexibility. Indeed, deoxygenation processes are able to handle FOG feeds with significantly higher free fatty acid concentrations relative to those typically required for biodiesel synthesis [11].

As shown in Scheme 1, FOG deoxygenation can proceed via hydrodeoxygenation (HDO) and decarboxylation/decarbonylation (deCOx). In HDO, oxygen is removed as H2O, and the alkanes produced have the same number of carbon atoms as the corresponding fatty acid chains comprising the FOG. In decarboxylation, oxygen is removed in the form of CO2, and in decarbonylation, oxygen is removed as H2O and CO. In both cases, the resulting alkanes have one carbon atom less than the corresponding fatty acid bound in the triglyceride. Depending on the catalyst and the experimental conditions employed, the CO and CO2 produced in the gas phase may react with hydrogen to form CH4. In fact, because of the number of confounding reactions, which include not only methanation but also the water gas shift (WGS) and the Boudouard reaction, these routes cannot be easily distinguished from each other based solely on the amounts of H2O, CO2, and CO produced [12,13].

**Scheme 1.** Deoxygenation routes for tristearin as a model compound representing triglycerides (blue shading) and concomitant reactions confounding oxygen-bearing deoxygenation products (red shading).

In recent years, the production of fuel-like hydrocarbons via deCOx has been intensively investigated as a way to avoid the large amounts and pressures of hydrogen, as well as the problematic sulfide catalysts required by HDO, since the hydrogen requirements of deCOx are lower and these reactions proceed over simple supported metal catalysts [14]. Although the majority of deCOx studies have focused on Pd and Pt catalysts, the high price of these metals has spurred the search for alternatives. Saliently, inexpensive Ni-based catalysts can provide comparable results to Pd and Pt formulations in the deCOx of FOG to hydrocarbons [15,16].

Since the high activity of Ni in C–C hydrogenolysis can decrease the carbon yield and the hydrogen e fficiency of deCOx processes, the incorporation of a second metal has been investigated as a means to modify the electronic and geometric properties of Ni to ultimately improve its activity and selectivity [17]. Indeed, Ni/Al2O3 promoted with Cu or Pt can a fford near quantitative diesel yields in the conversion of both model and realistic lipid feeds to fuel-like hydrocarbons [13,18]. The promotion effect displayed by these bimetallic catalysts is in large part attributed to the ability of Cu and Pt to facilitate NiO reduction at relatively low temperatures since metallic Ni sites constitute the active site for the deCOx reaction. Moreover, Pt addition also curbs the adsorption of CO on the catalyst surface, helping to avoid catalyst inhibition by any CO evolved via decarbonylation and the catalyst coking resulting from the disproportionation of CO via the Boudouard reaction [13].

Supported Ni catalysts promoted with Fe have also a fforded promising results in the conversion of model and realistic lipid feeds to hydrocarbons [17], the promoting e ffect of Fe being attributed to the synergy between nickel sites possessing the ability to activate hydrogen and iron sites with strong oxophilicity [17,19]. Indeed, since Fe has a higher oxygen a ffinity than Ni, oxygen vacancies within iron oxide species can facilitate the adsorption and subsequent activation of oxygenates. Specifically, H2 activated through its facile dissociative adsorption on Ni sites can spill over to neighboring Fe sites onto which the oxygen atoms of C=O groups are adsorbed, subsequent hydrogenation leading to deoxygenation products. In addition, the formation of Ni-Fe alloys with Fe-rich surfaces disrupts the adjacency of Ni atoms, a geometric e ffect known to suppress C–C hydrogenolysis, which requires Ni ensembles. Cu is also known to decrease the C–C hydrogenolysis activity of Ni through the same geometric e ffect [20].

In order to develop practicable catalytic deCOx technology for the conversion of FOG to fuel-like hydrocarbons, it is necessary to study the most promising catalysts using industrially relevant feeds and reaction conditions. In addition, in order to make meaningful comparisons between the performance of di fferent catalysts, measurements must be made at a steady state. Against this backdrop, the present work investigated the conversion of UCO to renewable diesel via deCOx over supported Ni catalysts promoted with Cu, Fe or Pt. The performance of these formulations was tested in a fixed bed reactor using industrially-relevant reaction conditions for 76 h of time on stream (TOS), as previous work has shown that catalysts of this type require >48 h of TOS to attain a steady state [11]. In addition, the analysis of the fresh, spent, and regenerated catalysts was undertaken in an e ffort to understand the distinct performance displayed by these formulations.
