Applied Catalysis B: Environmental 218 (2017) 430-442
`
`
`Contents lists available at ScienceDirect
`
` Applied Catalysis B: Environmental
`
`et is
`
`i
`
`
`
`journal homepage: www.elsevier.com/locate/apcatb
`ELSEVIER
`
`
`Co-oxidation of CO and propylene on Pd/CeO2-ZrO,z and Pd/Al,03
`monolith catalysts: A light-off, kinetics, and mechanistic study
`
`® CrossMark
`
`WendyLang, Paul Laing”, Yisun Cheng”, Carolyn Hubbard °,Michael P. Harold***
`4 Department of Chemical and Biomolecular Engineering, Texas Centerfor Clean Engines, Emissions & Fuels, University ofHouston, Houston, TX 77204,
`United States
`» Research and Innovation Center, Ford Motor Company, Dearborn, MI 48124, United States
`
`ARTICLE INFO
`
`ABSTRACT
`
`Article history:
`Received 27 February 2017
`Received in revised form 22 May 2017
`Accepted 20 June 2017
`Available online 21 June 2017
`K
`,
`P.thetum
`Ceria
`Three-waycatalyst
`Light-off
`Oxidation
`Carbon monoxide
`Propylene
`
`
`The light-off (ignition) and steady-state behavior for individual oxidation and co-oxidation of CO and
`propylene under near-stoichiometric conditions was studied using Pd/Alz03 and Pd/CeQ2-ZrO2 mono-
`lith catalysts. CO and propyleneare shown to be self- and mutually-inhibiting, with inhibition mitigated
`by the promotionaleffect of CeO2-ZrOz. Oxidation is enhanced at low temperatures for CO,andat inter-
`mediate and high temperatures for propylene, Light-off behavior during CO and propylene co-oxidation
`is similarly improved at low and high temperatures. Steady-state differential kinetics measurements
`using Pd/AlzO3 show reaction orders of ~—1 with respect to CO and propylene. Using Pd/CeO2-ZrO2, the
`reaction order with respect to CO shifts towards zero and the activation energy decreases, suggesting an
`alternate reaction mechanism for CO oxidation when enoughceria is present. Mechanisms for CO and
`propylene oxidation that are consistent with the kinetics and inhibition trends are presented.
`© 2017 Elsevier B.V. All rights reserved.
`
`1. Introduction
`
`Three-waycatalysts (TWCs)are used in gasoline engine vehicles
`to reduce exhaust emissions of carbon monoxide (CO), hydro-
`carbons (HCs), and nitrogen oxides (NOx). TWCs are capable of
`converting the three exhaust pollutants to products CO2, H20, and
`N2. As federally-mandated emission and fuel economy standards
`becomeincreasingly stringent, TWC light-off (LO) performance is
`critical in cost effectively meeting new and future regulations.
`Means of improving catalyst LO include catalyst modifications that
`minimizeinhibition effects of exhaust species and optimization of
`precious group metal (PGM)loading.
`TWCs consist of PGM dispersed on washcoated support mate-
`rials loaded on cordierite monoliths. Palladium (Pd) is the main
`oxidation componentin most modern automotive TWCs [1,2]. Its
`use in TWCsis incentivized by its superior thermal stability, a
`lowertendency to react with support materials (relative to base or
`non-noble metals), non-volatile oxides, and relatively lower cost
`and broaderavailability compared to Pt {1-3]. The type of cata-
`lyst support used affects the dispersion of precious metal sites,
`
`* Corresponding author.
`E-mail address: mharold@uh.edu (MLP. Harold).
`
`http://dx.doi.org/10.1016/j.apcatb.2017.06.064
`0926-3373/© 2017 Elsevier B.V. All rights reserved.
`
`thermal stability of components, surface area, pore volume, and
`surface reactivity [4]. Mixed oxides including alumina (Al203), ceria
`(CeO), and zirconia (ZrO2) are commonly used; Al2O3 provides a
`high surface area support andcarrier for precious metal catalysts
`[4,5]. AlgO3 further helps bind the catalyst layer to the substrate
`(e.g. cordierite) and also absorbs poisons, helping retain catalyst
`performance[3].
`To compensate for deviations from ideal stoichiometric three-
`wayCatalyst operating conditions, oxygen storage components are
`added to the catalyst, with ceria being the main oxygen storage
`componentusedtoday [1]. Ceria is incorporated in three-way cat-
`alysts becauseofits ability to be reduced and re-oxidized,cycling
`betweenthe bounds Ce(3+) (Ce203) and Ce(4+) (CeO2) to store oxy-
`gen, e.g. Ce203 + 0.503 — 2CeO2 [3,6,7]. This redox chemistry and
`oxygen storage capacity enables the oxidation of CO and HCs under
`rich transient conditions [3]. Ceria also has a stabilizing effect on
`Alz03 and precious metal dispersion [3,8,9]. Zirconia improves the
`activity and thermalstability of -y-Alz03 and ceria, and has been
`shown to prevent the grain growth ofceria crystallites at high
`temperatures[5].
`Pd/Ceria and Pd/Ceria-Zirconia (Pd/CZO) containing catalysts
`have been used and compared to conventional Pd/Al203 catalysts
`with promising results in ameliorating the self-inhibitory nature of
`CO and hydrocarbon oxidation [6-8,10-13]. LO under near ambient
`conditions is desired to reduce or overcome cold-start emissions;
`
`WRG-1013
`
`1
`
`WRG-1013
`
`

`

`W. Lang etal. / Applied Catalysis B: Environmental 218 (2017) 430-442
`
`431
`
`Table 1
`and, using Pd/Ceria powdercatalysts, LO of CO has been observed
`
`Catalysts synthesized.
`below 100°C [8,14]. Fromakinetics standpoint, incorporating ceria
`Fresh Catalyst
`Pd/CzZO
`Pd/Al2O3
`in PGM catalysts has been observed to increase the rate of CO
`wt% Pd in washcoat
`1.0
`1.0
`oxidation and reaction order with respect to CO relative to con-
`Pd loading (¢/ft?of monolith)
`55
`52
`ventional PGM/Al203 catalysts [6-10,11,15-17]. CO adsorbed on
`Washcoatloading (g/in? of monolith)
`3.2 g/in?, CZO
`3.0 g/in?, AlzO3
`and covering the PGM surface can react with not only adsorbed
`BET Surface Area (m2/g)
`69.4
`128
`and dissociated oxygen species (e.g. the oxidation mechanism on
`Pore width (A)
`326
`157
`PGM/Al203), but also oxygen supplied by the ceria support, i.e. via
`Pore volume (cm?/g)
`0.565
`0.504
`an alternate mechanism that improvesthe overall rate of reaction
`Nanoparticle size (A)
`865
`469
`% Dispersion
`11.4
`23.3
`{6-8,11,16-18]. This enhancementis attributed to the interfacial
`reaction between CO adsorbed on PGMactive sites and oxygen
`supplied by ceria in the catalyst support [8,11,18]. The degree of
`enhancementusing PGM/Ceria catalysts is dependenton the metal
`particle size (more specifically, the length of the interface between
`metalparticles and the ceria support), whereas alumina-based cat-
`alysts exhibit rates independentof particle size [11,18]. Saturation
`of PGMsites by COlikely limits the effects of size-dependence for
`PGM|/AI,0; catalysts; however, CO on the metal does not impact the
`rate of O2 adsorption onto ceria, and thus the higher the concen-
`tration of peripheralsites, the higher the rate of CO oxidation [11].
`Self-inhibition of other exhaust species such as propylene,acety-
`lene, and toluene have been observed using various PGMcatalysts
`{12,15,19-21]; and in the case of propylene, it has been shown
`that including ceria in Pt and Pd catalysts reducesself-inhibition
`and decreases LO temperatures relative to Pd/Al2O3 catalysts under
`stoichiometric conditions[13,20,22].
`While CO oxidation on PGM, PGM/Al,03, PGM/Ceria cata-
`lysts has been well-studied,
`the kinetics and mechanism of
`hydrocarbon oxidation and simultaneous oxidation of CO and
`hydrocarbons under application-relevant conditions using model
`Pd/AlpO3 versus Pd/Ceria monolith catalysts have seen far fewer
`studies. Further insights into optimizing the oxidative component
`of Pd-based TWCsare thus desired from the perspective of more
`comprehensively understanding self- and mutual-inhibition of CO
`and HCsandreaction kinetics and mechanisms within nonisother-
`
`2.1.2. Monolith washcoating
`Blank 400 cpsi cordierite monolith samples (BASF-NGK) 0.5 in
`diameterand length were used.Synthesized catalyst powders were
`mixed in deionized waterto create catalyst slurries. Dilute HNO3
`wasadded to reduce theslurry pH to approximately 3.5 for more
`effective ball milling and coating [23]. The catalyst slurries were
`ball-milled for 20h. Monoliths were washcoated by alternately
`dipping ends of monoliths into slurries. To remove excess slurry
`in channels and achieve uniform coatings, air was blown through
`each end of the monolith after each dip-coating. Monoliths were
`then dried at 120°C for 20h. This procedure was repeated sev-
`eral times as necessary to reach the desired washcoat loading,
`after which the monoliths were calcined at 500°C for 5h. Catalyst
`properties including loadings, BET surface areas, pore properties,
`nanoparticle size, and percentPd dispersion are provided in Table 1.
`Characterization was conducted using a Micromeritics ASAP 2020
`instrument: physisorption (BET, pore properties, nanoparticle size)
`was carried outusing N2 gas at 77 K and chemisorption (percent Pd
`dispersion) using H, gas at 35°C.
`
`2.2. Reactor testing
`
`mal monolith reactor systems.To this end,in this study, transient
`and steady-state reactor experiments using Pd monolith catalysts
`were conducted to moresystematically interprettherole ofsupport
`(CZO versus AlpO3) compositions on self-inhibition of individual
`species and co-inhibition during LO of CO and propylene (chosen
`as a model hydrocarbon). The key reactions studied and respective
`heats of reaction are:
`
`CO + 0.502-+ C02
`
`(AH, = —283kJ/molCO)
`
`C3Hg +4.502 > 3CO2+3H20
`
`(AH,= —1926kJ/molC3Hg)
`
`(1)
`
`(2)
`
`Reaction orders and activation energies of CO and propylene oxi-
`dation were determined and mechanistic reaction steps proposed.
`The results and postulated mechanisms provide insight into co-
`oxidation, inhibition, and optimization of catalyst composition and
`activity.
`
`2. Experimental
`
`2.1. Catalyst synthesis
`
`2.2.1, Reactor system
`The feed gas mixture was supplied by gas cylinders (Matheson
`Tri-Gas; Praxair, Inc.) and controlled using a series of mass flow
`controllers (MKS Inc.) Gas lines were wrapped with heating tape
`(heated to 130°C) to prevent water condensation within the sys-
`tem. The reactor consistedofa cylindrical quartz tube placed inside
`a Lindbergh temperature controlled furnace. The monolith catalyst
`was placed inside the reactor tube. Four thermocouples (Omega)
`were placed inside the reactor system to monitor temperatures at
`four different axial positions: 20 cm upstream of the monolith, 1 cm
`before the front face of the monolith to monitorthe feed (monolith
`entrance) gas temperature, halfway downthe length ofa central
`monolith channel to monitor the monolith temperature, and just
`inside the end of a monolith channel to measure the monolith end
`temperature. LabView™ software was used to control and record
`massflow controller and thermocouple signals. The primary ana-
`lytical system consisted of a FTIR spectrometer (MKS Inc., Multigas
`2030) used to detect concentrations of CO, C3Hs, and CO2. MKS
`software wasusedto record FTIR signals.
`
`2.1.1, Incipient wetness impregnation
`The catalysts were synthesized by incipient wetness impregna-
`tion. Pd/CZO and Pd/Al203 catalysts were prepared with 1 wt% Pd
`in the washcoat and anoverall washcoatloading of approximately
`3 g/in? of monolith. AlzO3 (Sasol, 140 m?/g) and CZO (Rhodia/Ford,
`70 m?/g) powders werecalcined in air at 500°C for 5h prior to
`impregnation. Pd was deposited onto the support powders via an
`aqueous solution of Pd(II) nitrate hydrate (Alfa Aesar). The impreg-
`nated powdercatalyst was dried at 120°C for 20 hand then calcined
`in air at 500°C for 5h.
`
`2.2.2, Experimental conditions
`Experiments were designed to simulate application-relevant
`space velocity (GHSV=150,000h-') with gas compositions com-
`prising 0.1-1% CO, 250-500 ppm C3Hg, O2, and balanceAr. To study
`the differential kinetics of oxidation (i.e. estimate reaction orders
`and activation energies), CO or C3Hg conversion was held below
`15%, with concentrations of one reactant varied while that of the
`other held constant. During LO experiments, 02 concentrations
`were adjusted to give an air-fuel equivalence ratio \ of 1.01 (i.e.
`
`2
`
`2
`
`

`

`432
`
`W.Lang et al. /Applied Catalysis B: Environmental 218 (2017) 430-442
`
`a near-stoichiometric, slightly lean mixture). \ is defined as the
`ratio of the actual air-to-fuel ratio (AFR) to the stoichiometric AFR:
`
`_ __AFR
`AFRgtoich
`
`(3)
`
`Xr
`
`Table 2 includes feed concentrations used in LO experiments.
`The feeds were devoid of H20 and COfor the current study. The
`lack of COis not expected to impact oxidation results. Though H20
`is noted to have an effect on CO and HC oxidation, it was excluded
`from feed mixtures to eliminate any complicating side reactions
`(e.g. water gas shift) and allow for focus on the fundamental per-
`formanceand kinetics of oxidation reactionsof interest.
`Catalysts were pretreated in 10% O2 (balance Ar) at 550°C
`for 30min before each set of experiments. For temperature
`programmed oxidation (TPO) experiments, the reactor furnace
`temperature was ramped at 3°C/min from room temperature
`to 500°C. Temperatures were held constant during steady-state
`experiments. The Pd/CZO catalyst was tested “fresh” (initial tests
`after synthesis) and after 10+ monthsofuse (“deactivated”).
`
`3. Results
`
`3.1. CO oxidation
`
`3.1.1. CO oxidation kinetics
`
`Steady-state kinetics experiments were conducted using the
`Pd/Al,03 and Pd/CZOcatalysts to estimate reaction orders andacti-
`vation energies. Conventional In(rate) versus In(Caye) plots give an
`estimate of the CO reaction order for a prescribed temperature,
`where Cave is the average of the reactant feed and effluent con-
`centrations. Fig. 1a and b showssuchdata for CO oxidation on the
`three catalysts at different monolith temperatures (T). The reac-
`tion order with respect to CO is ~—1 for the Pd/Al,O3 catalyst and
`~—0.52 to ~0.10 for the deactivated Pd/CZO catalyst. These obser-
`vations are consistent with literature values for PGM/Al,0O3 and
`PGM/CeO, catalysts, with negative reaction orders validating the
`self-inhibitory nature of CO and orders approaching zero demon-
`strating the promoting effects of ceria [7,15-18,24—26]. The error
`bounds shownin Fig. 1b lead to a range of CO reaction orders using
`Pd/CZO,Fig. 1c shows thatthe activation energy using the Pd/Al,03
`catalyst is in the range of 134-156 kJ/mol, consistent withliterature
`values | 15,25,26]. The observed activation energy decreasessignif-
`icantly to 25-70kJ/mol using the Pd/CZO catalyst. The activation
`energy for the Pd/Al203 catalyst correspondsto the binding energy
`of CO on Pd [15,25,26], while the much lowervaluefor the Pd/CZO
`catalyst suggests a change in reaction mechanism. Weelaborate
`on the mechanism in the discussionsection. Fig. 1d shows In(rate)
`plotted versus In(C;,,), where C;, is the inlet feed concentration of
`O2, The reaction order with respect to O2 using the Pd/Al2O3 cata-
`lyst is approximately 1, as reported in the literature for PGM/Al203
`catalysts [15,17,25-27]. Using the Pd/CZO catalyst, the observed
`reaction orderwith respect to O32 is slightly positive (approximately
`0.3); fractional orders (O-0.6) have been reported in theliterature
`for PGM/CeOz and PGM/CeO3/Alz03 catalysts [11,15,17,18].
`
`3.1.2. CO light-offbehavior
`Conversion versus monolith temperature results for CO TPO
`experiments are shown in Fig. 2, LO curves were generated as the
`reactor was ramped up in temperature while gas mixtures con-
`taining 0.5% or 1% CO werefed to the reactor. Using both Pd/Al203
`and Pd/CZOcatalysts, the temperature spanned a sufficiently wide
`range to include negligible conversion and nearly complete conver-
`sion. The Pd/CZO catalyst results are shown whenthecatalyst was
`tested fresh and deactivated. The data indicate that complete con-
`version ofCO is achieved with the required temperature dependent
`on the catalyst composition and CO concentration. With increasing
`
`CO feed concentration of CO, the LO curves shift to the right. The
`higher temperatures required to light-off the higher feed concen-
`trations of CO demonstrate that COis a self-inhibiting species. For
`example, using the Pd/Al,O3 catalyst, the monolith temperature
`at which 50% conversion of CO is achieved (T50) increases from
`216°C with a feed concentration of 0.5% CO to 225°C with 1% CO.
`Additional T50 temperatures are provided in Table 3. The apparent
`benefit ofceria is evident given that COlightoff occurs at lower tem-
`peratures using Pd/CZO, even as the Pd/CZO catalyst experiences a
`loss in activity with use overtime.As listed in Table 3,T10(monolith
`temperature at which 10% CO conversion is achieved) and T50 val-
`ues decrease using the Pd/CZO catalyst. Examining oxidation of 1%
`CO, T50 values are 225 °C with Pd/Al203, 164°C with fresh Pd/CZO,
`and 195 °C with deactivated Pd/CZO. The CO oxidationrateis clearly
`enhanced with Pd/CZO compared to Pd/Alz03.
`The activity of the deactivated Pd/CZO catalyst was further
`studied during steady-state CO oxidation experiments with feed
`concentrations of 0.5% CO, 0.6% CO, and 1% CO (A = 1.01 in all experi-
`ments). Steady-state conversion versus monolith temperature data
`are shownin Fig. 3. T10 and T50 temperatures increase with
`increasing CO concentration, demonstrating the self-inhibiting
`nature of CO under steady-state conditions as observed during
`transient temperature ramp experiments. Underdifferential, low
`conversion conditions, experimental error contributes to less clar-
`ity in the dependence of CO conversion and CO feed concentration.
`Weexplored this in the previous section on CO oxidation kinetics.
`In Fig. 4, monolith temperatures corresponding to Pd/CZO and
`Pd/Alz03 LO data in Fig. 1 are plotted versus gas feed temperature,
`which was recorded at the entrance of the monolith. CO oxida-
`tion is exothermic and the amountof heat released and thus the
`
`temperaturerise is dependent on CO concentration. A temperature
`rise is noted, the magnitude of whichis the difference between
`the monolith temperature and feed temperature. As expected, the
`temperature rise commencesat lower feed temperatures using the
`more active Pd/CZO catalyst for both studied feed concentrations
`of 0.5% and 1% CO. For both Pd/CZO and Pd/Al,O3 the maximum
`difference between monolith and feed temperatures during oxida-
`tion of 0.5% CO is approximately 65°C (feed temperature: 178°C),
`abouthalf of the roughly 120°C rise observed during 1% CO oxida-
`tion (feed temperature: 156°C). As the feed temperature increases,
`the difference between monolith and feed temperatures dimin-
`ishes slightly. For instance, using Pd/CZO to oxidize 1% CO,for a feed
`temperature of approximately 360°C, the monolith temperature is
`85°C higher. A comparison of the observed experimental temper-
`ature rise with theoretical adiabatic temperature rise is detailed in
`the discussion section.
`
`3.2. Propylene oxidation
`
`3.2.1. Propylene oxidation kinetics
`The steady-state propylene oxidation rate versus average con-
`centration data are plotted in Fig. 5a. The reaction order with
`respect to propylene is in the range of —0.78 to —0.75 for both
`Pd/Alz03 and Pd/CZO catalysts. The negative-orderis an expected
`outcome for Pd without ceria as propylene is a self-inhibiting
`species, with values comparable to literature values [15,28,29].
`Arrhenius plots shown in Fig. 5b reveal propylene oxidation activa-
`tion energies of 72-105 k]/molusing both catalysts, approximately
`half of that observed for CO oxidation and similarto literature val-
`
`ues [15,28]. Fig. 5c showsIn(rate) plotted versus In(C;,), where Cin
`is the inlet feed concentration of 02; reaction orders with respect
`to O2 using both catalysts are extrapolated from thefitted data. The
`reaction order with respect to O2 is in the range of 1.17-1.79 using
`the Pd/Al,O3 catalyst and 1.38-1.45 using the Pd/CZO catalyst,
`within range of literature values (0.2-1.5) for Pd based catalysts
`(Pd wire, Pd/SiO2, Pd/AlgO03, Pd/CeO2/Alz03) [15,28,29]. Table 4
`
`3
`
`3
`
`

`

`W. Lang et al, / Applied Catalysis B: Environmental 218 (2017) 430-442
`
`433
`
`Table 2
`
`LO experiment feed concentrations.
`Feed Gas
`
`COLO
`0.5
`
`0.45
`
`C3H, LO
`
`1.0
`
`0.70
`
`250
`0.31
`
`500
`0.42
`
`Mixture LO
`1.0
`500
`0.89
`
`n=-0.39
`T: 140.3 °C
`+0.9°C
`
`
`5
`|
`(a)
`ae
`n=o.10
`(b)
`r N= 0.02
`T 151.0°C
`L
`L T: 146.7°C +0.4°C
`08°C
`_
`b
`oO
`= dae |
`oO
`8
`L
`=
`=
`ei Pal
`=
`-
`
`L
`
`t
`
`|
`
`t
`
`-2.1
`
`Pd/AI,0,
`n=-0.94
`T: 200.2°C +1.3°C
`
`n=-0.52
`1 139.7°C £0.3°C
`[|
`— 6.5
`-2
`-1.9
`-1.8
`-2.1
`-1.9
`“17
`-1.5
`In(Caye (mol/L)
`In(Caye (Mol/L))
`
`-1.3
`
`| En
`Enact = 134 kJ/mol
`
`|
`
`2
`
`Palcz0
`
`Esct = 54 kJ/mol
`
`i
`
`i
`
`1
`2.1
`
`1
`
`1
`
`1
`2.2
`
`i
`
`1
`
`1
`2.3
`
`4000/T (K)
`
`-55
`
`:
`os ae |
`L
`o
`3so
`A.J
`L
`& -5.9 [
`
`= 61 |
`
`L
`L
`
`-6.3
`
`Pd/AI,O,
`aa
`T: 200.0 °C +1.3°C
`
`d
`(d)
`
`Pd/czo
`n=0.26
`T 145.6 °C +0.3°C
`
`24
`
`27
`
`2.5
`
`2.3
`
`-2.1
`
`-1.9
`
`-1.7
`
`-1.5
`
`In(C,, (mol/L))
`
`co (%)
`C3Hs (ppm)
`O2 (%)
`
`-5.45
`
`-5.6
`
`-5.75
`
`as
`
`oao £é=
`
`5.9
`
`5
`
`onH
`: -5.5
`
`5o E2o
`
`c=
`
`E=
`
`46
`
`6.5
`
`
`
`
`
`
`
`
`
`
`
`Fig. 1. (a) Reaction order with respect to CO on Pd/Al,03 (b) Reaction order with respect to CO on deactivated Pd/CZO (c) Activation energies of CO oxidation (0.6% CO) (d)
`Reaction orders with respect to 02. Vertical error bars and margin of error represent standard deviation from the mean.
`
`
`
`
`
`
`
`100
`+ PdiAlO,
`so} =--C0 0.5%
`7 —co1%
`80
`; Pdiczo
`| ---CO 0.5% }
`; —com |> fresh
`70
`| ===CO 0.5% }
`deactivated
`60
`[—coi% !
`So +
`
`40 -
`30 +
`20 +
`
`
`
`Conversion(%)
`
`Fig. 2. CO LO on Pd/Al,O, and Pd/CZO. Conditions: 0.5% or 1% CO, 0, for \ = 1.01, balance Ar, GHSV = 150,000 h-', temperature ramp 3°C/min.
`
`175
`Monolith Temperature (°C)
`
`225
`
`275
`
`325
`
`4
`
`

`

`434
`
`W. Lang et al. /Applied Catalysis B: Environmental 218 (2017) 430-442
`
`Table 3
`LO temperatures for CO oxidation.
`
`
`Catalyst
`CO feed
`
`T10 (°C)
`T50 (°C)
`Maximum conversion
`
`Pd/Alz03
`0.5%
`
`193
`216
`100
`
`1%
`
`209
`225
`100
`
`Pd/CZO (fresh, deactivated)
`0.5%
`
`95, 141
`157, 191
`100, 100
`
`1%
`
`127, 148
`164, 195
`100, 100
`
`—CO05%
`90 | —co0.6%
`ap|701%
`70 }
`
`100
`
`——
`
`125
`
`175
`Monolith Temperature (°C)
`
`225
`
`275
`
`325
`
`= 60 |
`50
`40 }
`30
`20 +
`10 +
`o
`
`$2
`
`$§
`
`oO
`
`
`
`
`
`
`Fig. 3. Steady-state CO oxidation on deactivated Pd/CZO. Conditions: 0.5%, 0.6%, or 1% CO, Oz for X=1.01, balance Ar, GHSV=150,000h~'. Vertical error bars represent
`standard error of the mean CO conversion.
`
`
`
`
`
`
`475 | —Feed (Monolith Entrance)
`r Pd/CZO
`400 L —Monolith, 1% CO
`---Monolith, 0.5% CO
`—
`2 305 [ Pd/AlO,
`»
`| ——Monolith, 1% CO
`=
`=-+Monolith, 0.5% CO
`© 250 |
`L
`
`a =
`
`@ 175
`
`100 +
`
`25
`
`25
`
`.
`100
`
`
`
`:
`325
`250
`175
`Feed (Monolith Entrance) Temperature (°C)
`
`400
`
`Fig. 4, Temperature rise during CO LO. Conditions: 0.5% or 1% CO, O2 for \=1,01, balance Ar, GHSV= 150,000 h-', temperature ramp 3°C/min.
`
`Table 4
`Kinetic parameters for CO, C;H, oxidation.
`
`Catalyst
`
`Pd/AlzO3
`Pd/czo
`
`CO Reaction
`Order
`-0.91
`—0,52 to 0.10
`
`CO Eat
`(kJ/mol)
`134-156
`25-70
`
`C3Hg, Reaction
`Order
`—0.78
`—0.75
`
`C3He Eat
`(kJ/mol)
`72-105
`88-95
`
`QO, Reaction Order
`for CO oxidation
`1.02-1.03
`0.26-0.29
`
`Oz Reaction Order
`for C3Hg oxidation
`1.17-1.79
`1.38-1.45
`
`lists rangesof reaction ordersand activation energies for additional
`data collected. Later we speculate on the nature of observed reac-
`tion orders and why C3Hg self-inhibition is not mitigated at low
`temperatures usingceria.
`
`3.2.2. Propylenelight-off behavior
`LO experiments similar to CO were carried out for propylene.
`Fig. 6 shows the LO behaviorofpropyleneusing feed concentrations
`of 250ppm and 500 ppm.Forboth catalysts, increasing the feed
`concentration of propyleneshifts the LO curves to higher tempera-
`tures. Thus, like CO, propylene is a self-inhibiting species. Whereas
`CO was completely converted under the experimental conditions,
`propylene was not, even at temperaturesas high as 500°C. As listed
`in Table 5, the asymptoticlimit of the propylene conversionis in the
`
`range of 93-94% using Pd/Al,03 and 98% using Pd/CZO. We expand
`on these limits and related masstransport effects in the discussion
`section. At lower temperatures (<200°C), propylene conversion is
`not enhanced with the additionof ceria to the extentthat it is dur-
`
`ing CO oxidation. The T10 temperaturesare fairly similar using both
`catalysts: roughly 170°C for a feed concentration of 250 ppm C3H,
`and 190°C for 500 ppm C3Hg. Another difference in comparison
`to CO oxidation is that over a range of intermediate temperatures
`(200-300°C), a lag in conversion or “shoulder” is observed for
`the LO curvesusing the Pd/Al,O3 catalyst. This unconventional LO
`curve feature (i.e., for a single overall catalytic reaction) does not
`appearduring propylene LO using Pd/CZO.As a result of the lag in
`conversion, T50 temperatures differ using the twocatalysts: for a
`feed concentration of 500 ppm C3Hg,, T50 values are 226°C using
`
`5
`
`5
`
`

`

`W. Lang etal. / Applied Catalysis B: Environmental 218 (2017) 430-442
`
`
`435
`
`ano
`
`|
`
`Pd/AI,0,
`n=-0.78
`T: 195.0°C+0.3°C
`
`(a)
`
`-7.8
`
`L
`Pr
`
`E.ct= 84 kJ/mol
`
`
`
`
`(b)
`
`=
`»
`$
`
`5 8.2
`3
`E
`2 8.4 |
`=
`66 |
`
`Pd/CZO
`Ent = 95 kJ/mol
`
`
`
`
` Pd/Al,O,
`
`
`
`
`3
`
`(%)866&@8888
` 250
`
`24 |
`o
`:
`
`/
`
`|
`
`8
`=
`Ee
`L
`g
`£88 +
`
`pdiczo
`n=-0.75
`| T193.6°C+03°C
`
`-39
`
`-5.1
`
`L
`-5
`
`1.
`
`1
`-4.9
`
`L
`
`L
`-4.8
`
`L
`
`1
`-4.7
`
`L
`
`1
`-4.6
`
`1
`
`L
`-4.5
`
`-4.4
`
`-8.8
`2.03
`
`2.04
`
`2.05
`
`2.06
`
`2.07
`
`2.08
`
`2.09
`
`2.1
`
`8
`
`In(C,,-(mol/L))
`
`Pd/CZO
`-62 | n=1.44
`T: 194.6°C+0.3°C
`
`
`
`(Cc)
`
`_
`
`1000/T (K)
`
`_
`
`a2
`
`o 8
`
`E2%
`
`=
`
`-8.4 +
`L
`O -86 |
`
`-88 |
`
`9
`
`Pd/AI,0,
`n=1.49
`T: 194.6°C +1.6°C
`
`9.2
`-2.75
`
`L— 1 1 ;
`-2.55
`-2.35
`
`-2.15
`
`In(C,,, (mol/L)
`
`Fig. 5. (a) Reaction orders with respect to C;Hg at 193-195°C (b) Activation energies of C;Hg oxidation using 500 ppm C3Hg, (c) Reaction orders with respect to O2 at
`195-199°C. Vertical error bars and margin of error represent standard deviation from the mean.
`
`
`
`
`8
`
`Pd/Al,O5
`| ---- C3H6 250 ppm
`— C3H6 500 ppm
`| Pdaiczo
`--=- C3H6 250 ppm
`— C3H6 500 ppm
`
`Conversion
`
`So
`
`25
`
`100
`
`Monolith Temperature (°C)
`
`475
`
`Fig. 6. C;Hg LO on Pd/Al203 and Pd/CZO. Conditions: 250 ppm or 500 ppm C3Hg, O02 for A = 1.01, balance Ar, GHSV= 150,000 h-', temperature ramp 3°C/min.
`
`Table 5
`LO temperatures for propylene oxidation.
`
`Catalyst
`C3Hg, feed
`
`T10(°C)
`T50 (°C)
`Maximum conversion
`
`Pd/Al203
`
`250 ppm
`172
`
`93
`
`500 ppm
`191
`270
`
`Pd/CZO
`
`250 ppm
`176
`208
`98
`
`500 ppm
`193
`226
`98
`
`6
`
`

`

`436
`
`W. Lang et al. /Applied Catalysis B: Environmental 218 (2017) 430-442
`
`Pd/CZO and 276°C using Pd/Al203. This trend may be interpreted
`as rate enhancementbyceria for propylene oxidation. We return
`to this pointlater.
`
`3.2.3. Propylene temperature ramp cycling vs. steady-state
`oxidation
`
`Using both Pd/Al203 and Pd/CZOcatalysts, additional temper-
`ature rampcycling and steady-state oxidation experiments were
`conducted using 500 ppmC3Hgin the feed (A=1.01). The intent
`was to obtain some additional information about the conversion
`
`shoulder and lack thereof observed using the Pd/Al,03 and Pd/CZO
`catalysts, respectively. Fig. 7 compares propylene conversion dur-
`ing steady-state oxidation and cycling experiments (during which
`the reactor is ramped up in temperature, then down,and up a sec-
`ond time) using (a) Pd/Al,03 and (b) Pd/CZO. Using bothcatalysts,
`with cycled temperature ramps, the oxidation temperaturesshift
`higher, demonstrating a clockwise hysteresis trend.
`Using the Pd/Al,03 catalyst, the shoulder is diminished with
`cycling. Furthermore, waiting even longer times to capture the
`actual steady-state (time invariant) oxidation trends, the conver-
`sion at any given monolith temperature is even lower and no
`shoulderat intermediate temperatures is apparent. T10 increases
`from 183°C during the first ramp-up to 201°C during both the
`ramp-down and second ramp-up to 216°C during steady-state
`oxidation. T50 is 233°C during the first ramp-up, 245°C during
`the ramp-down, 238°C during the second ramp-up, and 260°C at
`steady-state.
`The LO results for Pd/CZO show both similar and dissimilar
`features. As for Pd/Alz03, T10 temperatures increase over the
`course of cycling using the Pd/CZO catalyst; from 189°C during
`the first ramp-up to 205°C during the ramp-down. However, T10
`then shifts lower to 199°C during the second ramp-up and 203°C
`during steady-state oxidation. T50 shows a comparable trend;it
`increases from 223°C during the first ramp-up to 238°C during
`the ramp-down. During the second ramp-up, T50 is 224°C, i.e.
`the LO curveis similar to that of the first ramp-up above ~50%
`conversion, T50 during steady-state experiments is 233 °C - lower
`than that of the second ramp-up.In fact, the propylene conversion
`during steady-state oxidation is higher than that during the tem-
`perature ramp-down - contrary to what wasobserved using the
`Pd/Al203 catalyst. The conversion is lower during the ramp-down
`than the ramp-up’s, demonstrating clockwise hysteresis effects.
`While oxidationis inhibited at lower temperatures during the sec-
`ond ramp-up(suggesting inhibition due to carbonaceousspecies
`accumulating over the course of the cycling experiment), conver-
`sion matchesthat of the initial ramp-up. Ceria-supplied oxygen
`could play a greater role at intermediate and high temperatures
`over the course of the temperature ramp-up. During steady-state
`oxidation, the conversion is in fact higher than that during the
`temperature ramp-down.Allowing the catalyst to operate under
`the sameconditions overrelatively longer periods of time could
`lead to enhancementof re-oxidation of ceria and oxidation by ceria
`greater than that during the transient cooling down segmentof the
`cycling experiment. Mechanistic reaction steps are proposedin the
`discussion section to speculate the cause of these trends.
`
`3.3. CO+C3H¢ co-oxidation:light-off behavior
`
`3.3.1. Individual vs. co-oxidation
`
`alone to 266°C when1% COis oxidized in the presence of 500 ppm
`of propylene, demonstrating that propylene inhibits CO oxidation.
`While T50 for propylene oxidation doesnotshift significantly from
`the individual oxidation to co-oxidation, T10 increases from 191°C
`to 249°C,illustrating the inhibiting effect CO has on propylene oxi-
`dation. Thus, CO and propylene are mutually inhibiting. Comparing
`T10 temperaturesofall four LO curves in Fig. 8a, a change in order
`of LO is observed under these experimental conditions: 500 ppm
`propylene begins oxidizing at a lower temperature (191°C) than
`1% CO (209 °C) during individual oxidation, but during co-oxidation
`T10 for CO is 244°C and 249°C for propylene. An additional point
`of interest is the lack of a shoulder in conversion in the propylene
`LO curve during co-oxidation.
`Fig. 8b comparesthe analogousset of LO curves generated using
`the Pd/CZO catalyst. Under the same experimental conditions, CO
`and propyleneare similarly mutually inhibiting using the Pd/CZO
`catalyst. For CO oxidation, T10 increases by 53 °C and T50 increases
`by 65°C in the presence of 500 ppm propylene. Propylene oxida-
`tion LO temperaturesarealso listed in Table 6; both T10 and T50
`temperaturesincrease in the presence ofCO. Unlike oxidation order
`trends observed using Pd/Al203, CO lights off before propylene dur-
`ing both individual oxidation and co-oxidation using Pd/CZO, and
`no shoulderin LO curvesis observed.
`
`3.3.2. Co-oxidation compared using both catalysts
`Comparing the co-oxidation behavior of CO and propylene using
`both catalysts, Fig. 9 reiterates the observation that CO lights
`off before propylene under the experimental conditions studied.
`The lowest LO temperatures are achieved using the Pd/CZO cata-
`lyst. Another benefit of using CZO in the catalyst appears to have
`a sharper, more efficient ignition leading to higher conversion
`limits reached at lower temperatures relative to co-oxidation on
`Pd/Al203.
`
`4. Discussion
`
`4.1. CO oxidation mechanisms
`
`Pd-catalyzed CO oxidation is a well-studied reaction system,
`with many studies focused on the mechanism and/or kinetics
`[7,25,26,30-35]. Even more studies have appeared ontherelated
`Pt-catalyzed CO oxidation. Reaction mechanisms of CO oxidation
`using Pd on the two different support materials, AlpO3 and CZO,
`are depicted schematically in Fig. 10 [36].
`Pd/Alz03 catalyzed CO oxidationis generally consideredto fol-
`low the four-step sequence:
`
`cO+X=CO-X
`
`O02 +X =02-X
`
`02-X+X=20-X
`
`CO-X+0-X > CO2 + 2X
`
`(R1)
`
`(R2)
`
`(R3)
`
`(R4)
`
`where X denotes a single Pd site. Sometimes the mechanistic
`sequence is shownas three steps with steps R2 and R3 combined
`as follows:
`
`0) +2X = 20-X
`
`(R5)
`
`The co-oxidation behavior was examinedby co-feeding CO and
`propylene. In Fig. 8a, individual oxidation LO curves for 1% CO and
`500 ppm propylene are compared to curves generated during co-
`oxidation of a feed mixture of 1% CO and 500 ppm propylene using
`Pd/Al,03. When CO and propyleneare mixed and co-fed at the indi-
`cated concentrations, ignition is delayed to higher temperatures.
`As listed in Table 6, T50 shifts from 225°C when 1% COis oxidized
`
`Note that while (R5) is second order with respect to the vacantsites,
`the reaction is considered first-orderin the vacantsites.
`For temperatures below the LO temperature, CO adsorbs onto
`Pd active sites (X) dispersed on both support materials, preventing
`O2 from competitively adsorbing and dissociating. On Pd/Al2O3, the
`overall reaction is limited by CO desorption - as the temperature
`increases, CO desorption occurs and