How to Measure the Reaction Performance of Heterogeneous Catalytic Reactions Reliably

Ding Ma is a professor of Peking University. He is an advisory member for various journals, has been Associate Editor for the catalysis journal of RSC, Catalysis Science & Technology, since 2014, and was elected as a Fellow of the Royal Society of Chemistry in 2016. His research interests are heterogeneous catalysis, especially those related with energy issues, including C1 chemistry (methane, CO2, and syngas conversion), new reaction routes for sustainable chemistry, and the development of an in situ spectroscopic method that can be operated at working reaction conditions to study reaction mechanisms.Bingjun Xu is an Associate Professor in the Department of Chemical and Biomolecular Engineering at University of Delaware. He received his PhD in Physical Chemistry, advised by Prof. Friend, from Harvard University in 2011 and worked with Prof. Davis at Caltech as a postdoctoral researcher. The overarching goal of Xu research group is to develop innovative strategies to produce renewable electricity, fuels, and chemicals by the rational design of efficient thermo- and electro-catalytic processes. A special focus is placed on the mechanistic understanding of surface-mediated reactions by employing and developing state-of-the-art spectroscopic and kinetic techniques.Meng Wang is a Research Scientist in the College of Chemistry and Molecular Engineering at Peking University. He received his PhD in Chemistry from Technical University of Munich, advised by Prof. Lercher, in 2018 and worked in Pacific Northwest National Laboratory as a Research Associate from 2015 to 2019. Wang’s principal research task focuses on deciphering the kinetic and mechanistic pathways for catalytic reactions by combining reaction kinetics, isotopic experiments, and various in situ characterization methods.Mengtao Zhang is a PhD student in the College of Chemistry and Molecular Engineering at Peking University (Prof. Ding Ma). He received his BS (Chemistry) from Nanjing University in 2015. His research interests include water activation and hydrogen production, CO2 and other greenhouse gas conversion into desirable chemicals, design and development of new catalysts, and operando reaction mechanism and kinetics study. Ding Ma is a professor of Peking University. He is an advisory member for various journals, has been Associate Editor for the catalysis journal of RSC, Catalysis Science & Technology, since 2014, and was elected as a Fellow of the Royal Society of Chemistry in 2016. His research interests are heterogeneous catalysis, especially those related with energy issues, including C1 chemistry (methane, CO2, and syngas conversion), new reaction routes for sustainable chemistry, and the development of an in situ spectroscopic method that can be operated at working reaction conditions to study reaction mechanisms. Bingjun Xu is an Associate Professor in the Department of Chemical and Biomolecular Engineering at University of Delaware. He received his PhD in Physical Chemistry, advised by Prof. Friend, from Harvard University in 2011 and worked with Prof. Davis at Caltech as a postdoctoral researcher. The overarching goal of Xu research group is to develop innovative strategies to produce renewable electricity, fuels, and chemicals by the rational design of efficient thermo- and electro-catalytic processes. A special focus is placed on the mechanistic understanding of surface-mediated reactions by employing and developing state-of-the-art spectroscopic and kinetic techniques. Meng Wang is a Research Scientist in the College of Chemistry and Molecular Engineering at Peking University. He received his PhD in Chemistry from Technical University of Munich, advised by Prof. Lercher, in 2018 and worked in Pacific Northwest National Laboratory as a Research Associate from 2015 to 2019. Wang’s principal research task focuses on deciphering the kinetic and mechanistic pathways for catalytic reactions by combining reaction kinetics, isotopic experiments, and various in situ characterization methods. Mengtao Zhang is a PhD student in the College of Chemistry and Molecular Engineering at Peking University (Prof. Ding Ma). He received his BS (Chemistry) from Nanjing University in 2015. His research interests include water activation and hydrogen production, CO2 and other greenhouse gas conversion into desirable chemicals, design and development of new catalysts, and operando reaction mechanism and kinetics study. Heterogeneous catalysis, where the catalyst occupies a different phase from the reactants and products, is a central pillar in the modern chemicals and energy industries. The majority of chemicals and fuels are produced in heterogeneous catalytic processes, because of the extraordinary ability of catalytic sites to selectively accelerate targeted breaking and making of chemical bonds. Although thermal energy remains the dominant driving force in existing industrial catalytic processes, other emerging forms of energy, e.g., electron, photon, plasmon, and microwave, have been increasingly employed to drive heterogeneous catalysis. The development of active, selective, and durable catalysts is the engine in enabling green and cost-effective processes and thus has attracted much academic and industrial research effort in the design, synthesis, and characterization of catalytic materials with exquisite control of composition and structure. The structural complexity often makes the fair comparison of the performance across a wide range of catalytic materials challenging. Therefore, determining and reporting the intrinsic activity of catalysts on a per-site basis, i.e., the turnover frequency (TOF), in the kinetic regime is critical to the advancement of catalytic science.1Boudart M. Turnover Rates in Heterogeneous Catalysis.Chem. Rev. 1995; 95: 661-666Crossref Scopus (460) Google Scholar In this Commentary, we outline the fundamentals in the reliable measurement and analysis of reactivity and kinetic data, which hopefully could benefit researchers relatively new to the field of heterogeneous catalysis research. Accurate determination of key figures of merit in catalytic performance, e.g., the conversion, product distribution, and catalyst stability, is of utmost importance in the catalysis research. The reliability of reactivity data depends on many factors including reactor configuration, analytical methods, purity of the reactants, and experimental procedures and conditions, etc. The establishment of a rigorous mass balance of all elements (most commonly carbon, hydrogen, and oxygen) involved in the reaction system is an important measure of the quality of reactivity data, as any substantial deviation from the complete element balance, i.e., unity, will lead to errors in the calculations of most key figures of merit in catalytic performance. For example, with a poor carbon balance in the methane dehydroaromatization reaction over Mo/HZSM-5, the selectivity for benzene would be substantially overestimated without considering the carbon loss due to the coke formation on the catalyst.2Zheng H. Ma D. Liu X. Zhang W. Han X. Xu Y. Bao X. Methane dehydroaromatization over Mo/HZSM-5: A study of catalytic process.Catal. Lett. 2006; 111: 111-114Crossref Scopus (21) Google Scholar Element balance is typically determined by comparing the molar quantity of all species containing a specific element in the inputs (∑nin) and outputs (∑nout) of any given reaction system (Equation 1):Element balance = ∑nout∑nin× 100%= ∑nreactant−out+∑nproduct∑nreactant−in×100%.(Equation 1) ∑nin could be expressed as the molar sum of a specific element in the feed (∑nreactant-in), while ∑nout is the molar sum of this element in unreacted reactants (∑nreactant-out) and all products (∑nproduct). We note that all molar quantities in Equation 1 should be scaled with the number of atoms of the specific element in every species. It is essential to independently determine nreactant-out and ∑nproduct, as either value only provides an incomplete picture of the reaction system. For example, equating the sum of detected products with the amount of converted reactants implies the unsubstantiated assumption of a 100% element balance, which is likely the primary cause of unreliable results in some early methane aromatization studies.2Zheng H. Ma D. Liu X. Zhang W. Han X. Xu Y. Bao X. Methane dehydroaromatization over Mo/HZSM-5: A study of catalytic process.Catal. Lett. 2006; 111: 111-114Crossref Scopus (21) Google Scholar It is worth noting that close to unity (100%) element balance could be misleading at low conversions of a reactant, as the unreacted substrate artificially inflates the value. For example, for a reaction with a real conversion of 5%, a “decent” carbon balance of 97.5% is obtained using Equation 1 when only half of products are detected on a carbon basis. Although this reaction has a “high” carbon balance, the value calculated with Equation 1 is deceiving and most likely not representative at any substantial conversions at which practical catalytic processes operate. Thus, a modified element balance should be adopted in reactions with low conversions, e.g., <5%, which can be defined as the molar ratio between a specific element in products and the element in consumed reactants (Equation 2):modified element balance = ∑nproduced∑nconsumed×100%.(Equation 2) The choice of key elements in establishing quantitative balances depends on the nature of reactions. For example, hydrogen balance is as important as, if not more important than, carbon balance in the reforming reaction chemistry, while carbon balance is the undisputed primary concern in the Fischer-Tropsch synthesis (FTS), which is a well-established commercial catalytic process that converts a mixture of carbon monoxide and hydrogen into hydrocarbon. In addition to alkane and alkene formation, alcohols and other oxygenated hydrocarbons are also generated from competing reactions.3Zhai P. Controlling the selectivity in syngas conversion based on catalyst surface structure. Peking University, 2016Google Scholar The modified carbon balance expression (Equation 3) for FTS can be expressed as the sum of the selectivity of CO2 and hydrocarbons (HC) in the gas-phase analyzed online using a thermal conductivity detector (TCD) and a hydrogen flame ionization detector (FID), and the selectivity of products in the liquid phase, including HC with higher molecular weights and the oxygenates, collected from a cold trap and analyzed by offline gas chromatography (GC) after the oil-water separation (Figure 1A):Cbalance=nCO2+∑nHC−online+∑nHC−offline+∑noxygenate−offlinenCO−consumed×100% =Sel.CO2+∑(Sel.HC−online+Sel.HC−offline+Sel.oxygenate−offline)(Equation 3) In practice, deviation of carbon balance from unity could be caused by (but not limited to) carbonaceous deposition over the catalyst, loss of the products in offline collection, and evaporation of volatile products in the workup and transfer of the solution after reaction for batch reactors. Besides, the systematic error of chromatographic analysis could become the dominant source of error at differential conversions. Assuming perfect element balances in the calculation of selectivities is a practice to be avoided because it inflates the selectivities for detected products and thus leads to artificially more desirable results. For instance, normalizing the sum of selectivities for all detected products to 100% in a slurry-bed FTS reaction could be misleading given the possibility of undetected products and loss of products analyzed offline during collection. Actual carbon balances of <60% in FTS are not uncommon.3Zhai P. Controlling the selectivity in syngas conversion based on catalyst surface structure. Peking University, 2016Google Scholar Establishing a rigorous element balance is also critical and nontrivial when evaluating catalytic performances in closed systems, e.g., batch reactors, because the reaction could occur in multiple phases including liquid (aqueous and organic), gas, and solid. Thus, it is necessary to quantify products in each phase to ensure high element balances. For example, partition of products formed in the aqueous to the organic phase occurs during the reaction of vanillin over 5 wt% Pd/SWNT/SiO2 in a water-in-decalin emulsion in a batch reactor (Figure 1B).4Crossley S. Faria J. Shen M. Resasco D.E. Solid nanoparticles that catalyze biofuel upgrade reactions at the water/oil interface.Science. 2010; 327: 68-72Crossref PubMed Scopus (667) Google Scholar After reaction, the emulsion should be broken by filtering out the nanohybrid particles, followed by analyzing products in each phase to achieve high element balance. Despite the relatively high solubility of vanillin and vanillin alcohol in water, ∼20% of them are transferred into the decalin phase.4Crossley S. Faria J. Shen M. Resasco D.E. Solid nanoparticles that catalyze biofuel upgrade reactions at the water/oil interface.Science. 2010; 327: 68-72Crossref PubMed Scopus (667) Google Scholar Therefore, rigorous analysis of the reactants and products in both the aqueous and organic phases is critical. Accurate determination of the intrinsic catalyst activity is another key aspect in heterogeneous catalysis research. Any given heterogeneous catalytic reaction consists of several steps other than the chemical transformation(s) on the catalyst surface, including the interphase/intra-particle diffusion of reactants to and products from, as well as the adsorption of reactants to and the desorption of products from, the catalyst surface (Figure 2A). Interphase diffusion is the process in which reactants pass through the stationary layer containing reactants, products, and diluents, e.g., a carrier gas or solvent, to reach the external surface of the catalyst. For catalysts with a porous structure, the concentration gradient between the external and internal surfaces drives the intra-particle (pore) diffusion. Since the intrinsic activity of a catalyst refers exclusively to its ability to facilitate chemical transformations, catalysts must be evaluated at conditions under which rates are not impacted, let alone controlled, by mass and heat transport.5Madon R.J. Boudart M. Experimental criterion for the absence of artifacts in the measurement of rates of heterogeneous catalytic reactions.Ind. Eng. Chem. Fundam. 1982; 21: 438-447Crossref Scopus (388) Google Scholar Experimental techniques, such as non-isothermal integral reactor experiments, have been developed to obtain kinetic and transport information simultaneously, but these methods are challenging experimentally and time consuming in the data analysis. Alternatively, to ensure the measured activity reflects the intrinsic kinetics of the reaction occurring on the surface of the catalyst, several diagnostic criteria have been developed, such as Weisz-Prater criterion, Mears criterion, Bischoff criterion, etc.5Madon R.J. Boudart M. Experimental criterion for the absence of artifacts in the measurement of rates of heterogeneous catalytic reactions.Ind. Eng. Chem. Fundam. 1982; 21: 438-447Crossref Scopus (388) Google Scholar The well-known Madon-Boudart criterion below (Equation 4) is particularly useful in diagnosing whether interphase and/or intra-particle mass and heat transfer or both impact the measured rates:5Madon R.J. Boudart M. Experimental criterion for the absence of artifacts in the measurement of rates of heterogeneous catalytic reactions.Ind. Eng. Chem. Fundam. 1982; 21: 438-447Crossref Scopus (388) Google Scholarfr(mol s−1gcat.−1)molactivematerialgcat.−1=constant,(Equation 4) which states that the reaction rate (r) determined in the kinetic regime is directly proportional to the concentration of active material (f). In practice, the interphase diffusion limitation can be overcome by increasing the flow rate in continuous fixed-bed plug flow reactors at constant weight hourly space velocity (WHSV) until the reaction rate becomes a constant (Figure 2C),6Perego C. Peratello S. Experimental methods in catalytic kinetics.Catal. Today. 1999; 52: 133-145Crossref Scopus (217) Google Scholar because the intrinsic activity should be independent of the flow rate of reactants at a given WHSV according to the Madon-Boudart criterion. The particle size (R), which plays a major role in intra-particle diffusion, should not influence the rate either, so the most effective method of avoiding intra-particle diffusion in fixed-bed and batch reactors is to reduce the catalyst particle size (Figure 2D). For batch reactors (Figure 3A), aside from the particle size, the interphase diffusion can also be accelerated by increasing the agitation rate (Figure 3B, agitation test). For example, the interphase diffusion in FTS can safely be assumed to be minimal in typical industrial operations because of the high WHSV employed. The ratio between overall reaction rate and the intrinsic reaction rate, which is referred to as the effectiveness factor, approaches unity with decreasing particle diameter due to the reduced limitation of the intraparticle transport.7Shen W. Zhou J. Zhang B. INTRAPARTICLE DIFFUSION EFFECTS IN FISCHER-TROPSCH SYNTHESIS 1. MODELING OF DIFFUSION AND REACTION.J. Nat. Gas Chem. 1996; 5: 59-68Google Scholar Another common strategy to reduce interphase diffusion limitation is by diluting the catalyst bed with inert solids, typically quartz or SiC chips. This approach is especially helpful in strongly exothermic or endothermic reactions, which tend to induce a temperature gradient in the catalyst bed. Since the intrinsic rate is independent of the dilution, plotting measured rates as a function of the degree of dilution is a straightforward way to check whether interphase mass and heat transport have any impact on the kinetic measurements.Figure 3Transport Process and Intrinsic Activity in Batch ReactorShow full caption(A) Apparatus of the batch reactor.(B) Agitation test: the influence of rotation rate on conversion.(C) Concentration course: concentration changes of reactant A in batch reactor along with the reaction time (t). T0 and t1 indicates the starting point and the ending point of the reaction at elevated temperature, respectively.View Large Image Figure ViewerDownload Hi-res image Download (PPT) (A) Apparatus of the batch reactor. (B) Agitation test: the influence of rotation rate on conversion. (C) Concentration course: concentration changes of reactant A in batch reactor along with the reaction time (t). T0 and t1 indicates the starting point and the ending point of the reaction at elevated temperature, respectively. With the measured intrinsic rates, key kinetic parameters can be determined, e.g., rate constant (k), reaction orders (n), pre-exponential factor (A), and apparent activation energy (Ea) in Equations 5 and 6 by varying the reactant concentrations and temperature. Determination of those kinetic parameters is an accessible and reliable approach to evaluate the intrinsic activity of a catalyst and size reactors for scaling-up processes.rate=k[Reactant]n(Equation 5) k=Ae−Ea/RT(Equation 6) Methods for data analysis in extracting kinetic parameters depend on the type of reactor employed in the measurements. It is worth noting that well-defined correlations between measured rates and kinetic parameters only exist when reactions are conducted in one of the three types of ideal reactors, i.e., batch reactor, continuous stirred tank reactor (CSTR), and plug flow reactor (PFR). These mathematical correlations are typically introduced as “design equations” in most kinetics or reactor engineering textbooks. Although no real reactor is completely ideal, appropriate design equations could be used to determine the kinetic parameters if necessary precautions outlined below are taken. Reactors that lack an ideal flow pattern may occasionally be used to test catalysts but are not recommended for kinetic studies.6Perego C. Peratello S. Experimental methods in catalytic kinetics.Catal. Today. 1999; 52: 133-145Crossref Scopus (217) Google Scholar PFRs are widely used in gas-solid (G-S), liquid-solid (L-S), and gas-liquid-solid (G-L-S) systems. To approach ideal plug flow and prevent bypass in G-S and L-S systems, the inner diameter of reactor and the height of catalyst bed should not be too small compared to the catalyst particle size. Take Figure 2A as an example: the inner diameter (RI) should be at least 10× the particle size (R), and the bed height (h) to particle size (R) ratio should be larger than 50. PFRs for G-L-S systems are also referred to as trickle bed reactors, which are not effective for kinetic investigation due to the convoluted fluid dynamics and reaction kinetics. Isothermal and isobaric conditions should be maintained in reactivity and kinetic investigations in PFRs. A common practice to avoid temperature and pressure gradients is to operate the reactor at differential conversions (typically with the conversion of the reactants less than 5%), which is also commonly referred to as a differential reactor.6Perego C. Peratello S. Experimental methods in catalytic kinetics.Catal. Today. 1999; 52: 133-145Crossref Scopus (217) Google Scholar Despite the more challenging product analysis, operating at differential conversions has two key advantages. First, reaction rates can be directly calculated from the conversion under the condition of ultrashort residence time (Δτ→0); and second, reactors can be considered as gradient-less, meaning that temperature, pressure, and concentrations remain constant throughout the catalyst bed. In contrast, integral reactors operate at high conversions (Figure 2B), and it is necessary to tune the particle size of catalysts and introduce inert diluents in the catalyst bed for strongly exothermic or endothermic reactions to satisfy the isobaric and isothermal requirements, respectively. Further, determining intrinsic rates with an integral PFR reactor assumes the knowledge of spatial distribution of concenctrations of reactants, which are nontrivial to obtain. Comparing catalysts’ activities at high conversions should be avoided, as diffusion limitations could play a role in the performance. Stirred batch reactors (Figure 3A) also can be considered as integral reactors, and thus the initial rate should be calculated at the beginning of the reaction (t0). Special attention must be paid to the concentration of reactant(s) at t0, as some conversion could have occurred during the initial temperature ramp to the desired reaction temperature. Proper control experiments should be conducted to account for the conversion before reaction tempature is reached. Alternatively, separation of catalyst and reactants before the reaction is a practical method that minimizes the uncertainty in t0; e.g., storing the solid catalyst in a basket above the liquid reaction medium before its release could circumvent this challenge. Reactions with catalysts that require an induction period, or in situ activation, also deserve special treatments as the rate of reaction when the system just reaches the desired temperature may not be representative of the catalyst’s true performance. Again, control experiments are needed to determine the length of the induction period and the proper t0 (Figure 3C). Alternatively, introducing reactants from a separate container through a port after the in situ activation of catalyst in the solvent could reduce or eliminate the uncertainty of t0.4Crossley S. Faria J. Shen M. Resasco D.E. Solid nanoparticles that catalyze biofuel upgrade reactions at the water/oil interface.Science. 2010; 327: 68-72Crossref PubMed Scopus (667) Google Scholar Stirred tank reactors operate in the batch, semi-batch, or continuous mode (CSTR) and are also suitable for L-S, G-S, and G-L-S systems. As we have mentioned before, the agitation test can be used to ensure the interphase diffusion does not substantially impact the measured rates by conducting the experiments at different agitation rates (Figure 3B). In addition, L-S transport phenomena need to be examined separately by changing the amount of catalyst because the agitation has a less prominent impact on the L-S transport than the G-L transport in the G-L-S system. This type of reactor is widely used not only in thermocatalytic reactions with an integral heating/cooling system but also in electrocatalysis, photocatalysis, and other applications with different modifications. We note that when the reactor is customized for different aims, e.g., electrocatalysis and photocatalysis, the operating conditions must be optimized to ensure the reliability of the measured rates.8Weber R.S. Normalizing Hetereogeneous Electrocatalytic and Photocatalytic Rates.ACS Omega. 2019; 4: 4109-4112Crossref PubMed Scopus (6) Google Scholar Reactors are often assumed to be operating at the steady state without catalyst decays, i.e., the reaction rate will remain unchanged when the catalyst is reused in a batch reactor or at different reaction times in a flow reactor. It is always preferable to conduct kinetic analysis at conditions with a steady catalytic performance; however, catalyst deactivation is unavoidable in many hydrocarbon chemistries through a variety of pathways, such as poisoning, coking or fouling, sintering, and phase transformation. For example, poisoning by sulfur and nitrogen-containing contaminants and mechanical failure through attrition in slurry reactors are well-established deactivation mechanisms in the cobalt-catalyzed FTS.9Tsakoumis N.E. Rønning M. Borg Ø. Rytter E. Holmen A. Deactivation of cobalt based Fischer–Tropsch catalysts: A review.Catal. Today. 2010; 154: 162-182Crossref Scopus (471) Google Scholar In such cases, one common approach to address this challenge is to conduct multiple tests at increasingly lower conversions and extrapolate the rate at the zero conversion (r0) to determine intrinsic activities without the interference of deactivation. Catalyst stabilities are typically evaluated by plotting the activity and the product distribution as functions of time-on-stream (TOS) in PFRs and the number of times the catalyst is exposed to the reaction conditions in batch reactors. We note that evaluating catalyst stability at 100% conversion of the reactant should be avoided because the result could be heavily influenced by the catalyst loading. The best way to define the life of a catalyst is the number of turnovers (TON) a catalyst can produce.10Vannice M.A. Joyce W.H. Kinetics of Catalytic Reactions. Springer, 2005Crossref Scopus (357) Google Scholar When deactivation cannot be ignored in kinetic studies, measured reaction rates should be corrected. In PFRs, the impact of slight catalyst deactivation on the measured reaction rate (r) could be corrected by periodically returning to the same condition to quantify the degree of catalyst deactivation.11Forzatti P. Lietti L. Catalyst deactivation.Catal. Today. 1999; 52: 165-181Crossref Scopus (647) Google Scholar Evaluating and correcting the effect of catalyst deactivation in batch reactors are more involved because both deactivation and reaction kinetics can affect the variation of reactant concentration CA. Ngo et al. have established an effective method to assess catalyst deactivation in batch reactors: the evolution of reactant concentration CA is plotted as a function of the product of catalyst weight and time (w × t) for two different catalyst loadings. The curves overlap only in the absence of catalyst deactivation.12Ngo D.T. Sooknoi T. Resasco D.E. Improving stability of cyclopentanone aldol condensation MgO-based catalysts by surface hydrophobization with organosilanes.Appl. Catal. B. 2018; 237: 835-843Crossref Scopus (38) Google Scholar Chemical transformations in heterogeneous catalysis are generally considered to occur on specific locations on the catalyst surface, typically referred to as active sites, and intrinsic activity on a per site basis (i.e., TOF) offers a fair measure by which to compare catalysts with different composition and structure. TOF is calculated by dividing the intrinsic activity of a catalyst by its active site density, which typically has a unit of one over time (Equation 7), e.g., s−1. Intrinsic activity of catalysts can be normalized (and reported) with different quantities, e.g., overall mass, mass of a specific (expensive) component, and surface area, often to highlight certain features of a catalyst. Although justified in its own right, this practice makes the fair comparsion among different catalysts challenging. Meanwhile, TOF could be interpreted as the number of catalytic turnovers on a specific site in a given period of time, regardless of the composition and structure of the catalyst. Thus, TOF provides a reliable measure by which to compare the intrinsic activity of catalysts with different compositions and structures measured in different laboratories in a specific reaction.1Boudart M. Turnover Rates in Heterogeneous Catalysis.Chem. Rev. 1995; 95: 661-666Crossref Scopus (460) Google ScholarTOF=the number of reactant converted or product producedreaction time ×the number of active sites(Equation 7) It should be noted that one implicit assumption is made in the calculation of any TOF: for a catalyst, all the active sites on a catalyst have the same activity or one type of active site dominates the activity of a catalyst. With this assumption and the knowledge of the intrinsic activity of a catalyst (last section), the key to determine the TOF value is to identify the nature and quantify the density of active sites of the catalyst. We note that this is a process quite dependent on the nature of the reaction under study and often involves empirical or speculative assumptions. For example, the density of surface sites available to the reactant of a catalyst is frequently used to calculate TOFs. To obtain more accurate TOFs, it is crucial to quantify the density of active sites likely active in the reaction of interest. Selective titration with probe molecules, e.g., CO and hydrogen, via chemisorption could be a reliable method for structure-insensitive reactions. Alternatively, the amount of exposed metal atoms can be estimated by analyzing the size distribution of metal nanoparticles with electron microscopy when selective chemisorption is not feasible. It is important to note that these methods assume all exposed sites are equally active, which may not be true for many structure-sensitive or site-specific reactions. In reality, there is no compelling evidence supporting the assumption that all active sites on a catalyst have identical activity or that one type of active sites is dominant on a catalyst. Reaction pathways (hence, active sites) could change with experimental conditions, such as temperature and pressure. Thus, any TOF value normalized to a specific type of sites is an approximation at best. The heterogeneous nature of any solid surface means that multiple types of sites are present on any solid catalyst, e.g., different facets, edge, kink, and interfacial sites, and it is generally unlikely that they all possess the identical activity or drastically different activities such that one type of sites dominate. Thus, accurate determination of TOF requires titration methods capable of differentiating various surface sites, which are not generally available. Complementary spectroscopic techniques, such as IR and XAS, could be used to identify the nature of the active sites and determine the density of the surface sites. TOF determined by normalizing by all similar surface sites, e.g., metal sites quantified by CO chemisorption, generally underestimates the catalytic performance attributable to a specific type of active sites and should be regarded as the lower bound of the TOF, referred to as TOFmin. To obtain accurate TOF values for reactions, several specialized techniques including temporal analysis of products (TAP), steady-state isotopic-transient kinetic analysis (SSITKA), and modulation excitation spectroscopy have been developed to obtain so-called intrinsic TOF (or TOF maximum/TOFmax) and other kinetic and mechanistic information. TAP and SSITKA are suitable for the gas-solid reactions, while the modulation excitation spectroscopy method is designed to study catalytic reactions in the liquid phase. With the help of SSITKA, the apparent TOF values (TOFmin) can be deconvoluted to the intrinsic TOF (TOFmax), with determining the fraction of the surface active sites covered by reactive intermediates. The role of promoters, support, particle size, and process conditions can be deconvoluted. For example, Yang et al. found that Re promotion did not change the intrinsic TOF but slightly increased the coverage of reactive intermediates over carbon nanotube supported cobalt catalysts in the FTS.13Yang J. Chen D. Holmen A. Understanding the kinetics and Re promotion of carbon nanotube supported cobalt catalysts by SSITKA.Catal. Today. 2012; 186: 99-108Crossref Scopus (27) Google Scholar We refer interested readers to an excellent review on this topic by Davis14Davis R. Turnover rates on complex heterogeneous catalysts.AIChE J. 2018; 64: 3778-3785Crossref Scopus (16) Google Scholar Another practical consideration is that catalyst deactivation could reduce the density of active sites determined in ex situ measurements, which inflates the calculated TOF. Therefore, the TOF should be calculated either by rates when extrapolating to zero conversion with density of active site determined ex situ, or by rates determined at finite conversions with density of sites determined in situ or in operando. In summary, obtaining accurate and meaningful heterogeneous catalytic reaction data is crucial and nontrivial. In particular, the determination of key kinetic parameters, such as intrinsic reaction rate, TOF, and activation energy, require well-controlled experimental conditions and rigorous data analysis. Selecting an appropriate reactor configuration and reaction conditions (in the kinetic regime), and ensuring high elemental balances are crucial in obtaining high-quality data. Reporting TOFs at a given set of reaction conditions offers a straightforward way to compare data obtained over different catalysts and from different laboratories. Attention must be paid to determine accurate and meaningful TOF because it requires the accurate determination of intrinsic rates and counting of relevant active sites, both of which require careful experimental design and execution. We thank Prof. Maria Flytzani-Stephanopoulos and Prof. Daniel Resasco for the suggestions, discussion, and revision of the manuscript. This work was financially supported by Natural Science Foundation of China (21725301, 91645115, 21821004) and the National Key R&D Program of China (2017YFB0602200). B.X. acknowledges the support of the United States National Science Foundation CAREER Program (Award No. CBET-1744586).