F

UC

Ya

b

a

ARRAA

KCCNDI

1

lcoectif0i

PeetbAtt

C

h0

Applied Surface Science 403 (2017) 347–355

Contents lists available at ScienceDirect

Applied Surface Science

journa l homepage: www.e lsev ier .com/ locate /apsusc

ull Length Article

nderstanding the effect of CuO dispersion state on the activity ofuO modified Ce0.7Zr0.3O2 for NO removal

uan Cao a,b, Lianjun Liu a, Fei Gao a,b,∗, Lin Dong a,b,∗, Yi Chen a

Key Laboratory of Mesoscopic Chemistry of MOE, School of Chemistry and Chemical Engineering, Nanjing University, Nanjing 210093, PR ChinaJiangsu Key Laboratory of Vehicle Emissions Control, Center of Modern Analysis, Nanjing University, Nanjing 210093, PR China

r t i c l e i n f o

rticle history:eceived 28 July 2016eceived in revised form 19 January 2017ccepted 21 January 2017vailable online 22 January 2017

a b s t r a c t

In this work, CuO modified Ce0.7Zr0.3O2 (CZ) solid solution including CuO-deposited CZ (i.e., Cu/CZ) and−doped CZ (i.e., Cu-CZ) are prepared. The correlation of CuO neighboring structure with the catalytic per-formance for NO reduction have been proposed by various spectroscopic technologies. Results suggestedthat the strong synergetic effect between surface deposited copper species and CZ support can easilypromote the reducibility of Cu/CZ, which enhances its catalytic performance. In addition, the decomposi-

eywords:eO2-ZrO2 solid solutionuOO reductionispersion state

tion of NO occurs through different pathways over Cu/CZ and Cu-CZ, respectively, as suggested by in-situFTIR results. The distinct chemical state and environment of copper species between Cu/CZ and Cu-CZcan account for these differences. A surface model and the reaction mechanism are proposed to discussthe differences showing in the catalysts.

© 2017 Elsevier B.V. All rights reserved.

n-situ FTIR

. Introduction

Ceria (CeO2), with high oxygen storage capacity (OSC) and excel-ent redox behavior, has been widely adopted in the three-wayatalyst (TWC), and also applied as a support of metal oxidesr noble metal catalysts for CO oxidation [1,2]. However, it isspecially vulnerable to thermal sintering and the catalytic effi-iency easily decrease under reaction conditions. To overcomehese problems, numerous studies have been conducted to improvets performance by doping another element (e.g, Ti, Zr, Hf, La) toorm a solid solution. Among them, CexZr1-xO2, especially when.5 < x < 0.8, has been demonstrated to have an ideal performance

n OSC and catalytic activity.For traditional CexZr1-xO2 related catalysts, noble metals (e.g.,

t, Rh, Pd) were usually used as an active components [3]. How-ver, the high cost, low durability and poor resistance to poisoningnvironments motivated the community to explore an candidateo replace the noble metals. Recently, transition metal oxides haveeen intensively investigated as substitutions for noble metals.

mong them, copper oxide modified CexZr1-xO2 have been proven

o be promising for selective reduction of NO by CO and preferen-ial oxidation of CO [4–7], and many studies have been conducted

∗ Corresponding authors at: Jiangsu Key Laboratory of Vehicle Emissions Control,enter of Modern Analysis, Nanjing University, Nanjing 210093, PR China.

E-mail addresses: gaofei@nju.edu.cn (F. Gao), donglin@nju.edu.cn (L. Dong).

ttp://dx.doi.org/10.1016/j.apsusc.2017.01.212169-4332/© 2017 Elsevier B.V. All rights reserved.

to understand the materials structure/property, evaluate the cat-alytic performance, and explore the reaction mechanism over CuOmodified CexZr1-xO2. In our previous work [8], NO reduction byCO reaction was studied over CuO/CexZr1-xO2 (x = 0.2, 0.5 and 0.8)and the ceria rich catalyst CuO/Ce0.8Zr0.2O2 showed higher activitytoward NO reduction.

It was generally accepted that the performance of the CuO/CeO2catalyst was mainly influenced by the interaction between CuOand CeO2 support. The highly dispersed CuO species was provento be more active than bulk phase CuO [9]. However, Cu2+ ionsmay incorporate into CeO2 lattice to form a solid solutions at alow Cu/(Ce + Cu) ratio (<0.1) [10]. In this regard, a variety of copperoxide species (e.g, surface dispersed, lattice doped, and bulk CuO)may also exist at the interface of CuO and CexZr1-xO2. Hence, tohave a better understanding about it would be beneficial for thedesign and rationalization of the practical catalysts. In previousstudies, attempts have been made to study the role of CuO species,both incorporated into CexZr1-xO2 solid solutions, and dispersedon the surface of CexZr1-xO2 solid solutions. It was concluded thatfinely dispersed Cu2+ species have higher activity than those in thelattice of CeO2 or CeO2-ZrO2 [11–14]. Although a large amount ofstudies was carried out to study the structural and electronic prop-erties of CuO modified CexZr1-xO2 system, the interaction of copper

species with the neighboring environment has been frequently dis-missed, and the structural characteristics of copper species werestill a significant issue, which cannot be ruled out. The issue remainsambiguous and is open for discussion.

3 ce Sci

bmsvlncweretofrutt

cwXtioCtPs

2

2

pPwC2urawTc

tba

isao(

2

nisp

48 Y. Cao et al. / Applied Surfa

Moreover, the catalytic behavior of TWC can be greatly affectedy the nature of vehicle exhausts, which is directly deter-ined by the combustion conditions. Recently, lean-burn and

toichiometric-burn conditions are the most favorable scopes inehicle emission control. The rich-burn condition has drawn muchess attention on. Generally, noble metal catalysts are less vul-erable to reducing atmosphere. The valance state of the activeenters was relatively stable in rich burn conditions. However,hen it comes to transition metal oxide catalysts, the reductive

nvironment can have a major effect on their existing status. Sinceich-burn conditions are not negligible in an operating vehicle,specially at the cold-start stage. We believe it is of significanceo consider the effect of reductive atmosphere on the performancef transition metal oxide catalysts, which will be helpful rein-orce a more comprehensive understanding on the TWC. Hence, theeaction mechanism and the effect of the different copper speciesnder reductive atmosphere (for instance, NO + CO reaction withhe molar ratio of NO and CO is 1:2) need a better understanding athe atomic level.

In this present work, CuO-Ce0.7Zr0.3O2 and CuO/Ce0.7Zr0.3O2atalysts were prepared by surfactant-assistant hydrothermal andet-impregnation methods, respectively, and characterized byRD, BET, Raman, ICP, HR-TEM, EPR, H2-TPR, and in-situ FTIR

echnologies. An appropriate NO + CO model reaction under reduc-ng condition (NO:CO = 1:2, mol ratio) was also applied. The mainbjective of the present work was to (1) explore the correlation ofuO neighboring environment and the catalytic property, (2) inves-igate the adsorption behaviors of NO and/or CO on these catalysts.ossible reaction models were also proposed to try to establish thetructure-performance relationship.

. Experimental

.1. Catalysts preparation

Ce0.7Zr0.3O2 solid solution (denoted as CZ, thereafter) wasrepared by a hydrothermal co-precipitation method. 0.20 mmol123 (Tergitol(t.m.)xh(nonionic)) was dissolved in 25 ml distilledater, then mixed with an ionic solution containing 7.0 mmol

e(NO3)3·6H2O and 3.0 mmol ZrCl2O·8H2O in 5 ml distilled water.00 mmol of urea was added drop wise into the mixture in 2 hnder continuous stirring, then aged at 80 ◦C for 72 h in oven. Theesulting mixtures were transferred to Teflon-lined stainless steelutoclaves, maintained at 120 ◦C for 48 h. The slurries were filtered,ashed and dried, then calcined at 450 ◦C for 3 h in flowing air.

he resulting products were used as supports for copper-basedatalysts.

Copper doped CZ, labeled as Cu-CZ, was synthesized followinghe above procedure except that the ionic solution was preparedy dissolving 1.0 mmol Cu(NO3)2·3H2O, 7.0 mmol Ce(NO3)3·6H2Ond 3.0 mmol ZrCl2O·8H2O in 5 ml distilled water.

Copper supported CZ, denoted as Cu/CZ, was prepared by incip-ent wetness impregnation of CZ, with aqueous copper nitrateolutions. The resulting samples were dried overnight and calcinedt 450 ◦C for 3 h in flowing air afterwards. The Cu metal conetntf Cu-CZ and Cu/CZ were analyzed by inductively coupled plasmaICP) and found to be 0.6 wt%.

.2. Catalyst characterization

Surface areas of these supports and catalysts were measured by

itrogen adsorption at 77 K with BET method, by using Micromerit-

cs ASAP-2020 adsorption apparatus. The concentrations of copperpecies in these samples were determined by Inductively Cou-led Plasma (ICP). Known amount of samples were dissolved in

ence 403 (2017) 347–355

the mixture of nitro acid and hydrofluoric acid, water heated at100 ◦C to remove hydrofluoric acid. The solutions were analyzedin J-A1100 plasma spectrometer after dilution. XRD patterns werecollected on a Philips X’pert Pro diffractometer by Cu K� radia-tion (�=0.15418 nm). The mean grain sizes were calculated by theScherrer equation D = K�/�cos�. Lattice parameters were calcu-lated by Rietveld refinement, the Fm3m space group was assumed.Raman spectra were obtained on a Jobin-Yvon T64000 type LaserRaman spectroscopy, with 300 mW laser power and an excitedwavelength at 532 nm. UV–vis DRS spectra were recorded in therange of 200–900 nm by a UV-vis-NIR 5000 spectrophotometer.EPR measurements were performed on a Bruker EMX-10/12 spec-trometer operating at X-band frequency (m ≈ 9.4 GHz) and 100 kHzfiled modulation. The spectra of these catalysts were recorded at110 K. Temperature-programmed reduction (TPR) was carried outin a quartz U-tube reactor using 50 mg sample for each measure-ment. The sample was pretreated at 100 ◦C in nitrogen stream for1 h. After that, TPR started from room temperature at a rate of10 ◦C min−1 in a H2-Ar stream (7% H2 by volume). In situ FTIR spec-tra were collected from 400 to 4000 cm−1 at a resolution 4 cm−1

(number of scans = 32) on a Nicolet 5700 FTIR spectrometer. Athin self-supporting disc (about 25 mg catalyst) were prepared andmounted in an IR cell. It was therefore pretreated for 1 h at 300 ◦C inflowing N2. After cooling to room temperature, the self-supportingdisc were exposed to a controlled stream of CO-Ar (10% CO) or/andNO-Ar (5% NO) at a rate of 35.0 ml·min−1 for 30 min. In situ FTIRspectra (temperature-programmed reduction and reaction) wererecorded at ascending target temperatures (every 10 ◦C from 50 ◦Cto 300 ◦C at a rate of 5 ◦C min−1) by subtraction of appropriate back-ground reference.

2.3. Catalytic activity tests

The activities of the catalysts were determined under light-offprocedure, involving a feed steam with a fixed composition, 2.5%NO, 5% CO and 92.5% He by volume as diluents. The catalysts (50 mg)were pretreated in N2 stream at 300 ◦C for 1 h and then cooledto room temperature, after that, the mixed gases were switchedon. The reactions were carried out at different temperatures witha space velocity of 12,000 mL g−1 h−1. Two gas chromatographsequipped with thermal conduction detections were used for ana-lyzing the production. Column A with Paropak Q for separatingN2O and CO2 and column B packed with 5A and 13X moleculesieve (40–60 M) for separating N2, NO and CO. Values of percentageconversion and production yield are defined as follows:

XNO = FinNO − Fout

NO

FinNO

× 100%

YN2O =2Fout

N2O

FinNO

× 100%

YN2 =2Fout

N2

FinNO

× 100%

Where X and Y are the percentage conversion and productionyield, respectively. F is the molar flow of the inlet or outlet gas.

Kinetic measurements were performed under differential reac-tion conditions, with an appropriate amount of catalyst (5–30 mg)to limit the conversion of NO between 5% and 20%. The apparentactivation energy was calculated by Arrhenius equation.

r = A exp( Ea / RT) (1)

Where “r” stands for the reaction rate, “A” for the pre-exponential factor, “Ea” for the apparent activation energy. The

Y. Cao et al. / Applied Surface Science 403 (2017) 347–355 349

aNto

r

r

or

3

3

asccotlrsssidCwtnz

stsdo6fp

Fig. 2. Raman spectra for CZ, Cu-CZ and Cu/CZ samples.

Fig. 1. XRD patterns for CZ, Cu-CZ, Cu/CZ and pure CeO2.

pparent activation energy of NO conversion to N2O (Ea[N2O]) andO conversion to N2 (Ea[N2O]) were measured at different tempera-

ure, respectively. The reaction rate of NO conversion to N2O (r[N2O])r N2 (r[N2]) was acquired by the following equations.

[N2O] = 2N[N2O]/Wcat (2)

[N2] = 2N[N2]/Wcat (3)

N[N2O]andN[N2]are the N2O and N2 molar gas flow rate in theutlet gas, respectively. Wcat is the catalyst weight. The reactionate r[N2O] and r[N2]was in mol g−1 s−1.

. Results and discussions

.1. Dispersion state of CuO species on Ce0.7Zr0.3O2

XRD has been conducted to approach the structure of CZ beforend after CuO loading, as shown in Fig. 1a. The Cu/CZ and Cu-CZamples displayed a same cubic fluorite structure as CZ support. Norystalline copper oxides or metallic copper were observed, indi-ating CuO species were either highly dispersed on the CZ surfacer doped into the CZ lattice or beyond the XRD detection limita-ion. To gain insights into the dispersion state of CuO species, theattice constants of CZ, Cu/CZ and Cu-CZ were calculated by Rietveldefinement for comparison. As shown in Table 1, Cu/CZ displayed aimilar lattice constant as pure CZ support (0.537 nm), while Cu-CZhowed a slightly smaller lattice constant (0.535 nm) than CZ. Thislight lattice shrinkage might indicate that a fraction of bulk ceriumons or zirconium ions in Cu-CZ have been displaced by Cu2+ ionsue to the smaller ionic radius (Ce4+ = 0.097 nm, Zr4+ = 0.084 nm,u2+ = 0.073 nm) [8,15]. On the other hand, copper species in Cu/CZere mainly dispersed on the surface of CZ. It should be noted

hat the lattice constants of CZ and Cu/CZ were below that of CeO2anoparticles (0.541 nm), which confirms the formation of ceria-irconia solid solution [16].

Raman spectroscopy, a sensitive tool to characterize the surfacetructure of metal oxides, has been carried out to further approachhe dispersion state of copper species and the structure of CZ. Ashown in Fig. 2, a main band near 460 cm−1, standing for the triplyegenerate F2g mode of cubic fluorite ceria-zirconia [17,18] was

bserved on all samples. Two additional bands at 263 cm−1 and01 cm−1 are linked to oxygen vacancies (Vo) in ceria-zirconia sur-ace or subsurface [19]. A broad band at 1173 cm−1 is ascribed to therimary A1g asymmetry of CeO2 [20]. To find the effect of CuO in/on

Fig. 3. EPR spectra for Cu/CZ and Cu-CZ samples.

CZ, the full width at half maximum (FWHM) of the F2g peak was cal-culated and shown in Table 1. Clearly, the FWHM was in the order ofCu/CZ > Cu-CZ > CZ. Since the FWHM is considered mostly affectedby grain size and lattice disorder [21], it can be inferred that surfacecopper species probably create more defects in CZ compared withthe doped ones as the grain sizes were similar over these three sam-ples. The peak area ratios of 601 and 1173 cm−1 bands to 460 cm−1

were also calculated for comparison. The ratio of A610/A460 (R1)could represent for the Vo concentration, and A1173/A460 (R2) maystand for the degree of lattice disorder [11], and these area ratioswere shown in Table 1. It can be seen that the R1 and R2 of bothCu/CZ and Cu-CZ have several times higher than that of pure CZsample. Moreover, the R1 and R2 of Cu/CZ are higher than these ofCu-CZ sample. Again, this result demonstrated that more surfaceVo and defects were induced by surface dispersed CuO species, andthe doped CuO species may be located in the bulk phase.

To further examine the existing status of copper species, Cu/CZ

and Cu-CZ were characterized by EPR spectroscopy, as shown inFig. 3. The spin densities are determined by using copper sulfate asa standard [22]. The spin active copper species in Cu-CZ are about

350 Y. Cao et al. / Applied Surface Science 403 (2017) 347–355

Table 1Surface area, Grain size, Lattice constant and FWHM of Raman spectra for CZ supports and copper based catalysts.

samples Surface area (m2 g−1) Grain size (nm) Lattice constant (nm) A610/A460 (R1) A1173/A460 (R2) FWHM of F2g Peak(cm−1)

0.08 0.07 410.15 0.14 530.01 0.02 31

6EpiCCdEtAtCtaatsmrfs(Ctwgw

cpmCC

3

Csio2lpabnbapt

CmCr2t

Cu-CZ 74.2 8.3 0.535

Cu/CZ 84.6 9.5 0.537

CZ 86.3 8.8 0.537

9%, and that in Cu/CZ are about 66%. Since Cu-O-Cu networks arePR silent due to the magnetic interaction between neighboringaramagnetic species. It can be concluded that most copper species

n Cu/CZ and Cu-CZ are highly dispersed Cu2+ species. For bothu/CZ and Cu-CZ, four hyperfine splitting features for paramagneticu2+ ions (3d9) appeared on the spectra in the low field region. Toistinguish the coordination structure of Cu2+, the parameters ofPR signal were calculated and shown in Fig. 3. On can easily foundhat the spin parameters for Cu2+ species on Cu-CZ were g// = 2.308,// = 131 G, g⊥ = 2.037, A⊥ = 23G, which could be attributed to lat-

ice incorporated Cu2+ in Cu-CZ. By contrast, the parameters onu/CZ were g// = 2.350, A// = 120 G, g⊥ = 2.037, which were relatedo the surface modified Cu2+ in Cu/CZ [22–24]. This result reason-bly demonstrated that Cu2+ in Cu-CZ and Cu/CZ were indeed in

different chemical environment. In Fig. 3, it also can be foundhat the hyperfine splitting features of Cu-CZ were narrower andharper than that of Cu/CZ. As the surface environment was muchore complicated than bulk, the heterogeneity of chemical envi-

onment might result in the broadening of the hyperfine splittingeatures in Cu/CZ. Another important finding in Fig. 3 was that aignal in an axial symmetry at high field were observed in Cu/CZg‖ = 1.968 g⊥ = 1.946), which was attributed to the paramagnetice3+ ion in distorted octahedral symmetry [12]. On the contrary,his signal for Ce3+ was negligible in the Cu-CZ. This observationas in accordance with Raman results (Table 1) that more Ce3+ was

enerated in Cu/CZ than Cu-CZ for charge balance because more Voere produced in Cu/CZ.

The existence of copper species in these catalysts has also beenharacterized by UV-vis-DRS (Fig. S1). In comparison with CZ sup-ort, a d-d transfer band of Cu2+ is clearly observed in copperodified catalysts. The centers of d-d transfer band are 686 nm for

u-CZ and 719 nm for Cu/CZ, which emphasized their difference inu2+ coordination states [12].

.2. Reduction properties

H2-TPR was conducted to study the reduction properties ofu/CZ and Cu-CZ, and the profiles were shown in Fig. 4. Pure CZhowed a broad peak that started from 300 ◦C and reached the max-mum intensity at 521 ◦C, which was due to the reduction of surfacexygen of CZ. For Cu/CZ, two strong reduction peaks appeared at07 and 247 ◦C. While for Cu-CZ, these two reduction peaks were

ocated at a higher temperature range of 283 and 406 ◦C. The twoeaks were attributed to the concurrent reduction of copper speciesnd surface oxygen of CZ, probably due to the strong interactionetween Cu2+ and CZ support [8,11,25]. Notably, Cu/CZ showedot only much lower reduction temperature than that of Cu-CZ,ut also a larger H2 consumption amount for the reduction of Cu2+

nd surface oxygen. This difference suggested that the surface dis-ersed copper oxide species have a stronger synergetic effect withhe support in comparison with the lattice doped ones.

To further explore the reduction of CuO species in Cu/CZ and Cu-Z, in situ FTIR was performed in a TPR procedure with CO as a probeolecule. Fig. 5 shows the in situ FTIR spectra of CO interaction with

u/CZ and Cu-CZ. As shown in Fig. 5a, exposure of Cu/CZ to CO nearoom temperature did not give IR respond between 1900 cm−1 and200 cm−1, possibly due to the denominated Cu2+ on the surfacehat showed a weak interaction with CO. [26] Increasing the tem-

Fig. 4. H2-TPR profiles for CZ support and copper based catalysts.

perature led to an appearance of a new band at about 2106 cm−1 forCu+-carbonyl (Cu+-CO) [27], As shown in Fig. 5c, the band reachedthe maximum intensity at 120 ◦C. The increase of Cu+-CO peak waslikely due to the reduction of Cu2+ to Cu+, which again demon-strated that the initial surface CuO species were Cu2+ ions. Furtherincreasing the temperature to 300 ◦C induced the decrement of Cu+-CO, which was possibly caused either by thermal desorption or afurther reduction of Cu+ to Cu0, with respect to the fact that theabsorption band of Cu0-CO was too weak to be discovered. [26]Similar evolution behaviors of Cu+-CO species were also observedon Cu-CZ (Fig. 5b). It can be found in Fig. 5c that the Cu+-CO on Cu-CZreached its maximum intensity at a higher temperature (150 ◦C) incomparison with Cu/CZ. Moreover, a new band at 2169 cm−1, whichwas due to the adsorption of CO on Ce3+ ions [28], emerged above250 ◦C on Cu/CZ. By Contrast, this 2169 cm−1 band was almost invis-ible on Cu-CZ. By combining with the H2-TPR results, the differencesin FTIR results (i.e., lower temperature for the formation/desorptionof Cu+-CO and the appearance of Ce3+ on Cu/CZ) suggested thatthe dispersion state of CuO played a crucial role to determine theirreduction property. Surface dispersed copper species could rapidlychange their valance and obviously activate surface oxygen of CZ inthis TPR procedure, while lattice incorporated ones were difficultto be totally reduced to Cu0 below 300 ◦C either by CO or by H2.

An incorporation model was proposed to establish the corre-lation of the reduction property with the structure of Cu-CZ andCu/CZ. In consistent with our previous studies and the literaturereports, the preferential exposed plane of CZ was (111) plane,where surface octahedral vacancies were exposed [29,30]. As evi-denced by XRD, Raman and EPR results, Cu2+ in Cu/CZ was dispersedon the surface, which could occupy the surface octahedral vacanciesof the (111) plane with a capping oxygen to keep electro neutrality(Fig. 6a). It could form an asymmetrical five-coordination structurein which the capping oxygen was highly exposed [8]. This struc-ture was not stable and the coordination of capping oxygen was

unsaturated, and thus inducing the easier reduction of Cu2+ andsurface oxygen. On the other hand, Cu2+ ions incorporated into thefluorite lattice of CZ and replaced small amounts of cerium ions

Y. Cao et al. / Applied Surface Science 403 (2017) 347–355 351

Fig. 5. In situ FTIR spectra of CO adsorption on (a) Cu/CZ (bolded curve: 120 ◦C) and (b) Cu-CZ (bolded curve: 150 ◦C) at different temperatures. The vertical axis representsabsorbance in absorbance units (a.u.). (c) Integrated intensity of IR spectra for Cu+-CO over Cu-CZ and Cu/CZ.

ronme

(ts

3

sseeCtFiftyNv

Fig. 6. Proposed surface models and neighboring envi

Fig. 6b). Unlike the former five-coordination structure, this struc-ure appears difficult to be reduced since it is coordination andymmetric stable.

.3. Catalytic performance in NO + CO model reaction

Fig. 7 showed NO conversions, N2O and N2 yields over theseamples as a function of temperature. NO conversion over the CZupport can be ignored below 250 ◦C, and it was no more than 20%ven at 400 ◦C. After the introduction of CuO species, the samplesxhibited well reactivity in NO + CO model reaction. However, Cu-Z and Cu/CZ showed different catalytic performances, indicatinghe dispersion state of CuO species affects the catalytic activity. Inig. 7b, the N2O was favored during the initial stage (below 250 ◦C),t reached the maximum N2O yield near 225 ◦C for Cu-CZ and 175 ◦Cor Cu/CZ, where the N2 yield was lower than 10%. While at high

emperature region (250 ◦C for Cu/CZ and 300 ◦C for Cu-CZ), N2Oield would be decreased with the increasing temperature, where2 was the main product. Cu/CZ showed obviously higher NO con-ersion and N2 yield than Cu-CZ.

nts of copper ion species for (a) Cu/CZ and (b) Cu-CZ.

The existence of copper species in these catalysts after thetreatment in reaction condition has also been characterized. Thetreatment was carried on as described in the supporting infor-mation. The UV-vis-DRS results of these catalysts before and afterreaction are given in Fig. S1. The d-d bands of Cu2+ have a slightlyblue shift for Cu/CZ, and almost kept unchanged for Cu-CZ. EPRspectra (Fig. S2) indicate that for both Cu-CZ and Cu/CZ, thehyperfine splitting features of Cu2+ have not obviously changedafter the treatment by reaction condition, implying the existingstatus of copper species are relatively stable. Noticeably, the para-magnetic signal for Ce3+ (g‖ = 1.968, g⊥ = 1.946) is vanished afterthe treatment by NO + CO mixture. Moreover, EPR active Cu2+

species in Cu/CZ have obviously decreased after the treatment.Probably due to the electron transfer between Cu2+ and Ce3+

(Ce3+ + Cu2+ → Ce4+ + Cu+) induced by the reaction atmosphere.To further gain insight into the catalytic behavior toward those

catalysts, the kinetics of the reaction was studied. The apparent

activation energy of NO conversion to N2O (Ea[N2O]) was measuredbetween 160 ◦C and 230 ◦C. At which temperature, N2 selectivitywas relatively low and the reaction 2NO + CO → N2O +CO2 took thedominating place. As shown in Fig. 8, the Ea[N2O] of Cu-CZ and Cu/CZ

352 Y. Cao et al. / Applied Surface Science 403 (2017) 347–355

Fc

sfNtast

dabfanasrs

pdCpetc

3

N

ig. 7. NO + CO activity test results for CZ support and copper based catalysts (a) NOonversion; (b) N2O yield (solid lines) and N2 yield (dotted lines).

amples was 46.8 kJ mol−1 and 26.8 kJ mol−1, respectively. The dif-erent activation energy suggested that the reaction pathway ofO conversion to N2O on these catalysts was quite different. Since

he composition, the crystalline size and BET surface area of Cu-CZnd Cu/CZ were similar, the coordination environment of copperpecies should bear the main responsibility for their different reac-ion behavior.

At higher temperature, N2 selectivity raised and the completeecomposition of NO to N2 was the main reaction. The apparentctivation energy of NO conversion to N2 (Ea[N2]) was measuredetween 270 ◦C and 400 ◦C. Although these samples showed dif-erent reaction rate, their apparent activation energy Ea[N2]differs

little, near 120 kJ mol−1. It suggested that the reaction mecha-ism at higher temperature was probably the same toward Cu-CZnd Cu/CZ. The reaction may take place on the reactive sites by theame pathway. However, Cu/CZ showed obviously higher reactionate of NO conversion to N2 than Cu-CZ, indicating more reactiveites on Cu/CZ.

Associated with other characterizations, the grain sizes, cop-er concentrations and surface areas of these catalysts seem toraw minor effect to the difference of catalytic activity betweenu-CZ and Cu/CZ. Therefore, the chemical environment and dis-ersion state of copper oxide species determined the synergeticffect between copper and supports. The unstable five coordina-ion structure of Cu2+ in Cu/CZ should be responsible for its betteratalytic performance.

.4. NO + CO co-adsorption

In order to investigate the nature and the mechanism ofO reduction, the co-adsorption of NO and CO on those sam-

Fig. 8. Arrhenius plot for copper based catalysts (a) NO conversion to N2O; (b) NOconversion to N2.

ples was performed to imitate the environment during reaction(Fig. 9). NO molecules preferentially interact with these samplesnear room temperature. The spectra were similar with sole NOadsorption between 1000 and 1800 cm−1 (Fig. S3), with bandscorresponding to differently coordinated nitro/nitrite species.According to literatures, those bands could be assigned to chelatingnitro (1243–1245 cm−1) [31], monodentate nitrates (1278–1294,1538 cm−1), [8,32] bidenate nitrates (1589 cm−1), [33] and bridgenitrates (1027–1041, 1207–1213, 1619–1631 cm−1) [7,8,34]. Onthe other hand, there is no obvious band in the region between1900 cm−1 and 2280 cm−1 at room temperature. Further increas-ing the temperature did not have direct impact on the adsorbedspecies, before it comes to the breaking point. When it comes tothe breaking point (bolded curve in Fig. 9, 180 ◦C for Cu-CZ and140 ◦C for Cu/CZ), the bands stand for bridge nitrates decreasedswiftly. Meanwhile, there appeared two new bands centered at1294 cm−1 and 1074–1078 cm−1, which could be assigned to car-bonate species [35,36]. Gaseous N2O (2240 cm−1, 2208 cm−1) andCu+-CO (2106 cm−1) also emerged at that point. The formation ofN2O could not be simply originated from the direct decompositionof nitrates, since N2O was not detected in sole NO adsorption pro-cess (data not shown). These results indicated that reaction of theadsorbed nitrate species with CO had readily taken place ratherthan their rearrangement or desorption. When temperature wasabove the breaking point, bands in the region 1000–1800 cm−1

+

hardly changes. However, Cu -CO and gaseous N2O graduallydecreases with the increasing temperature, accompanied with thegradual increase of a broad band near 2170 cm−1 (Fig. 9(a, c)). Itwas much broader and more intensified than the Ce3+-CO band

Y. Cao et al. / Applied Surface Science 403 (2017) 347–355 353

F e: 140a

iiiss2sdcact

ig. 9. In situ FTIR spectra of CO and NO co adsorption on (a), (b) Cu/CZ (bolded curvxis represents absorbance in absorbance units (a.u.).

n sole CO adsorption. For sole NO adsorption, there is no signaln the region between 1900 cm−1 and 2280 cm−1. Consequently, itmplied that the broad band was contributed by a species relatedimultaneously with NO and CO. It could be assigned to isocyanatepecies adsorbed on different sites. The band at 2173 cm−1 and154 cm−1 are assigned to −NCO species attached to two Ce3+

ites differing in their coordinative saturation. The higher oxygeneficiency near cationic center leads to stronger Ce3+-NCO bond,onsequently, the asymmetric stretching vibration of −NCO should

ppear at lower frequency [3]. The higher reduction degree of Cu/CZould lead to more anion vacancies, which was responsible forhe presence of the band at 2154 cm−1. Another band emerged

Scheme 1. Schematic stepwise decomposition of NO over Cu-CZ a

◦C) and (c), (d) Cu-CZ (bolded curve: 180 ◦C) at different temperatures. The vertical

at 2192 cm−1 in Cu/CZ is assigned to Cu-NCO according to litera-tures [37]. This suggested highly exposed copper species in Cu/CZprovided additional adsorption sites for NCO group.

In an early attempt to explain CO oxidation in the presence ofNO, the reduction of NO in the presence of CO was assumed to pro-ceed in two steps. The first step is the partial reduction of NO toN2O, which was favored at lower temperature [38]. As discussedabove, the appearance of N2O was accompanied with the forma-tion of Cu+-CO species and the decomposition of bridge nitrites.

It is reasonable to deduce that Cu+ or Cu+-CO species was essen-tial for the partial reduction of bridge nitrites. The copper ions inCu/CZ exposed itself in higher degree than Cu-CZ. Meanwhile, cop-

nd Cu/CZ at (a) lower temperature (b) higher temperature.

3 ce Sci

pbbOf1ntNfsTTtss

pTs1ttedsritrwbtptoaCwuwdCte

am4ptattgadfHCmtagl

[[[

[

[[[[[

[

[[

[

54 Y. Cao et al. / Applied Surfa

er oxide species in Cu/CZ was easier to be reduced as sustainedy H2-TPR and CO-IR results. The quick change of copper valenceetween Cu2+ and Cu+ contributed to the higher activity of Cu/CZ.n the other hand, the positions of bridge nitrites (Fig. 9) were dif-

erent between Cu/CZ (1619, 1207, 1027 cm−1) and Cu-CZ (1631,213, 1041 cm−1). According to our previous work [8], the waveumber shift suggested that bridge nitrites have stronger interac-ion with Cu/CZ, so that N O bond was weakened and the adsorbedOx species on Cu/CZ were relatively unstable. It was responsible

or the lower apparent activation energy (Ea[N2O]) of NO conver-ion to N2O over Cu/CZ (26.8 kJ mol−1) than Cu-CZ (46.8 kJ mol−1).he nearby surface copper species could account for this effect.hese findings demonstrated that the difference in surface struc-ure determines the redox property of copper species, and also thetability of adsorbed nitrite species. The reaction pathway could bechematized according to Scheme 1.

The N2 yield started to be higher than N2O yield when the tem-erature was above 250 ◦C for Cu/CZ and 300 ◦C for Cu-CZ (Fig. 7b).he main reaction was the reduction of NO to N2. These sampleshowed almost the same apparent activation energy Ea[N2] (near20 kJ mol−1) at this step despite their different surface struc-ures. Their reaction pathway should be similar with each othero have parallel apparent activation energy. According to the lit-ratures, NCO was suggested to be the principal intermediateuring the second step [4,39–41]. The formation and decompo-ition procedure of adsorbed NCO was the crucial part for theeaction. NCO species were observed on both Cu-CZ and Cu/CZ,mplying the reaction had readily taken place. However, the reac-ion rate of Cu/CZ was obviously larger than Cu-CZ, implying moreeactive sites on Cu/CZ with respect to the similar reaction path-ay between these catalysts. Cu/CZ has a strong synergetic effect

etween ceria and surface cuprum ions. It enhances the chargeransfer and accelerates the generation of oxygen vacancies. Theresence of Ce3+-NCO species with lower lattice oxygen satura-ion (2154 cm−1) could emphasize the fact: there is vast numberf oxygen vacancies in Cu/CZ at reaction conditions, showing wellccordance with H2-TPR, LRS and EPR results. On the other hand,u/CZ showed both Cu-NCO (2192 cm−1) and Ce3+-NCO speciesith lower (2154 cm−1) and higher (2170 cm−1) lattice oxygen sat-

ration in NO + CO co-adsorption, while Cu-CZ showed Ce3+-NCOith higher lattice oxygen saturation only. It implied more surface

efects and active sites on Cu/CZ, which enhanced its reaction rate.onsequently, Cu/CZ showed better performance in NO reductiono N2 while Cu-CZ and Cu/CZ share the same apparent activationnergy (Ea[N2]).

To further prove the above conclusions, Cu-CZ and Cu/CZ cat-lysts were testified in NO + CO model reaction in the reversedanner, by stepwise decreasing the reaction temperature from

00 ◦C to 100 ◦C. The reversed catalytic activities of these sam-les were given in Fig. S4. The results showed accordance withhe proposed stepwise mechanism in Scheme 1. At high temper-ture region (above 250 ◦C), the N2 yield of Cu/CZ is still higherhan Cu-CZ. Noticeably, the N2 yield is higher than that in regularemperature-increasing test for both Cu/CZ and Cu-CZ. The oxy-en vacancies have readily generated in the reductive atmospheret higher temperatures. It could well support the formation andecomposition of NCO species. On the other hand, the N2O yield

or Cu/CZ is slightly higher than Cu-CZ in the reversed activity test.owever, they are obviously lower than that in regular test for bothu-CZ and Cu/CZ. As shown in Fig. 9(b,d), carbonates are one of theajor adsorbates above 200 ◦C. During the temperature-decreasing

est, the pre-adsorbed carbonates can hinder the adsorption of nitro

nd nitrates. It matched with the proposed pathway that N2O isenerated by the partial reduction of surface adsorbed nitrites atower temperature.

[

[

ence 403 (2017) 347–355

4. Conclusions

This work studied the different effects of copper dispersion stateon the activity of NO reduction. The results indicate that temper-ature and the coordination states of copper species determine thedecomposition pathways of NO species. The decomposition of sur-face adsorbed nitrates take place at lower temperature, the productwas mainly N2O. At higher temperature, the NCO species was theprincipal intermediate to generate N2. The dispersion states of CuOspecies have an important role in both reactions. Surface modifiedcopper oxide species in Cu/CZ could form an unstable, asymmetri-cal five-coordination structure. H2-TPR and CO-IR result revealedthat the synergetic effect between surface modified copper speciesand CZ support could easily promote the reducibility of Cu2+, whichenhanced the catalytic performance of Cu/CZ. CuO species in Cu-CZ introduced into the CZ fluorite structure, Cu2+ ions substitutea small portion of Ce4+ or Zr4+, form a relatively simplified andstabilized structure, thus these Cu2+ were relatively difficult to bereduced. These factors can explain the superior activity of Cu/CZtowards Cu-CZ.

Acknowledgments

The financial supports of the National Natural Science Foun-dation of China (No. 21573105), Natural Science Foundation ofJiangsu Province (BK20161392), and Jiangsu Province Science andTechnology Support Program(Industrial, BE2014130) are gratefullyacknowledged.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.apsusc.2017.01.212.

References

[1] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.[2] N. Kakuta, S. Ikawa, T. Eguchi, K. Murakami, H. Ohkita, T. Mizushima, J. Alloys.

Compd. 408 (2006) 1078.[3] M. Kantcheva, O. Samarskaya, L. Ilieva, G. Pantaleo, A.M. Venezia, D. Andreeva,

Appl. Catal. B: Environ. 88 (2009) 113.[4] G. Avgouropoulos, T. Ioannides, Appl. Catal. B: Environ. 67 (2006) 1.[5] M. Manzoli, R. Di Monte, F. Boccuzzi, S. Coluccia, J. Kaspar, Appl. Catal. B:

Environ. 61 (2005) 192.[6] S.P. Wang, T.Y. Zhang, Y. Su, S.R. Wang, S.M. Zhang, B.L. Zhu, S.H. Wu, Catal.

Lett. 121 (2008) 70.[7] L.J. Liu, B. Liu, L.H. Dong, J. Zhu, H.Q. Wan, K.Q. Sun, H.Y. Zhu, L. Dong, Y. Chen,

Appl. Catal. B: Environ. 90 (2009) 578.[8] L.J. Liu, Z.J. Yao, B. Liu, L. Dong, J. Catal. 275 (2010) 45.[9] W. Liu, M. Flytzani-Stephanopoulos, J. Catal. 153 (1995) 304.10] W.J. Shan, W.J. Shen, C. Li, Chem. Mater. 15 (2003) 4761.11] A.P. Jia, G.S. Hu, L. Meng, Y.L. Xie, J.Q. Lu, M.F. Luo, J. Catal. 289 (2012) 199.12] P. Ratnasamy, D. Srinivas, C.V.V. Satyanarayana, P. Manikandan, R.S.

SenthilKumaran, M. Sachin, V. Shetti, J. Catal. 221 (2004) 455.13] D.S. Zhang, Y.L. Qian, L.Y. Shi, H.L. Mai, R.H. Gao, J.P. Zhang, W.J. Yu, W.G. Cao,

Catal. Commun. 26 (2012) 164.14] M.F. Luo, J.M. Ma, J.Q. Lu, Y.P. Song, Y.J. Wang, J. Catal. 246 (2007) 52.15] W.J. Shan, Z.C. Feng, Z.L. Li, J. Zhang, W.J. Shen, C. Li, J. Catal. 228 (2004) 206.16] J. Kaspar, P. Fornasiero, M. Graziani, Catal. Today 50 (1999) 285.17] W.H. Weber, K.C. Hass, J.R. McBride, Phys. Rev. B 48 (1993) 178.18] J.E. Spanier, R.D. Robinson, F. Zhang, S.-W. Chan, I.P. Herman, Phys. Rev. B 64

(2001) 245407.19] J.R. McBride, K.C. Hass, B.D. Poindexter, W.H. Weber, J. Appl. Phys. 76 (1994)

2435.20] S. Velu, K. Suzuki, T. Osaki, Catal. Lett. 62 (1999) 159.21] M.D. Hernández-Alonso, A.B. Hungría, A. Martínez-Arias, J.M. Coronado, J.C.

Conesa, J. Soria, M. Fernández-García, Phys. Chem. Chem. Phys. 6 (2004) 3524.22] A. Martínez-Arias, M. Fernández-García, A.B. Hungría, A. Iglesias- Juez, O.

Gálvez, J.A. Anderson, J.C. Conesa, J. Soria, G. Munuera, J. Catal. 214 (2003) 261.23] P.G. Harrison, I.K. Ball, W. Azelee, W. Daniell, D. Goldfarb, Chem. Mater. 12

(2000) 3715.24] A. Aboukais, A. Bennani, C.F. Aïssi, G. Wrobel, M. Guelton, J.C. Vedrine, J. Chem.

Soc. Faraday Trans. 88 (1992) 615.

ce Sci

[

[[

[[

[

[[

[

[

[[

[

Y. Cao et al. / Applied Surfa

25] D. Gamarra, A. Hornés, Zs. Koppány, Z. Schay, G. Munuera, J. Soria, A.Martínez-Arias, J. Power Sources 169 (2007) 110.

26] K.I. Hadjiivanov, G.N. Vayssilov, Adv. Catal. 47 (2002) 307.27] R.D. Monte, J. Kaspar, P. Fornasiero, M. Graziani, C. Pazé, G. Gubitosa, Inorg.

Chim. Acta 334 (2002) 318.28] A. Badri, C. Binet, J.C. Lavalley, J. Chem. Soc. Faraday Trans. 92 (1996) 1603.29] L.J. Liu, Q. Yu, J. Zhu, H.Q. Wan, K.Q. Sun, B. Liu, H.Y. Zhu, F. Gao, L. Dong, Y.

Chen, J. Colloid and Interf. Sci. 349 (2010) 246.

30] J. Zhu, L.L. Zhang, Y. Deng, B. Liu, L.H. Dong, F. Gao, K.Q. Sun, L. Dong, Y. Chen,

Appl. Catal. B: Environ. 96 (2010) 449.31] Y.W. Chi, S.S.C. Chuang, J. Catal. 190 (2000) 75.32] B. Azambre, L. Zenboury, A. Koch, J.V. Weber, J. Phys. Chem. C 113 (2009)

13287.

[[[[

ence 403 (2017) 347–355 355

33] K. Shimizu, H. Kawabata, H. Maeshima, A. Satsuma, T. Hattori, J. Phys. Chem. B104 (2000) 2885.

34] I. Atribak, B. Azambre, A.B. Lopez, A. García-García, Appl. Catal. B: Environ. 92(2009) 126.

35] C. Morterra, G. Magnacca, Catal. Today 27 (1996) 497.36] L.J. Liu, Y. Cao, W.J. Sun, Z.J. Yao, B. Liu, F. Gao, L. Dong, Catal. Today 175 (2011)

48.37] F. Solymosi, L. Völgyesi, J. Sárkány, J. Catal. 54 (1978) 336.

38] M. Shelef, J.T. Kummer, Chem. Eng. Prog. Symp. Ser. 67 (1972) 115.39] W.C. Hecker, A.T. Bell, J. Catal. 84 (1983) 200.40] Y. Ukisu, S. Sato, A. Abe, K. Yoshida, Appl. Catal. B:Environ. 2 (1993) 147.41] E.Z. Ciftlikli, J. Lallo, E.Y.-M. Lee, S. Rangan, S.D. Senanayake, B.J. Hinch, J. Catal.

295 (2012) 269.