CO2-fuel 光催化反应-掺碘TIO2

Applied Catalysis A:General 400 (2011) 195–202Contents lists available at ScienceDirectApplied Catalysis A:Generalj o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /a p c a taVisible light responsive iodine-doped TiO 2for photocatalytic reduction of CO 2to fuelsQianyi Zhang a ,Ying Li a ,∗,Erik A.Ackerman b ,Marija Gajdardziska-Josifovska c ,Hailong Li daUniversity of Wisconsin-Milwaukee,Department of Mechanical Engineering,Milwaukee,WI 53211,USA bUniversity of Wisconsin-Milwaukee,Department of Electrical Engineering,Milwaukee,WI 53211,USA cUniversity of Wisconsin-Milwaukee,Department of Physics and Laboratory for Surface Studies,Milwaukee,WI 53211,USA dUniversity of Florida,Department of Environmental Engineering and Sciences,Gainesville,FL 32611,USAa r t i c l e i n f o Article history:Received 28February 2011Received in revised form 21April 2011Accepted 25April 2011Available online 3 May 2011Keywords:PhotocatalysisSolar energy conversion TiO 2Nanocomposite CO 2reductiona b s t r a c tIodine-doped titanium oxide (I-TiO 2)nanoparticles that are photocatalitically responsive to visible light illumination have been synthesized by hydrothermal method.The structure and properties of I-TiO 2nanocrystals prepared with different iodine doping levels and/or calcination temperatures were char-acterized by X-ray diffraction,transmission electron microscopy and diffraction,X-ray photoelectron spectroscopy,and UV–vis diffuse reflectance spectra.The three nominal iodine dopant levels (5,10,15wt.%)and the two lower calcination temperatures (375,450◦C)produced mixture of anatase and brookite nanocrystals,with small fraction of rutile found at 550◦C.The anatase phase of TiO 2increased in volume fraction with increased calcination temperature and iodine levels.The photocatalytic activi-ties of the I-TiO 2powders were investigated by photocatalytic reduction of CO 2with H 2O under visible light ( >400nm)and also under UV–vis illumination.CO was found to be the major photoreduction product using both undoped and doped TiO 2.A high CO 2reduction activity was observed for I-TiO 2cata-lysts (highest CO yield equivalent to 2.4␮mol g −1h −1)under visible light,and they also had much higher CO 2photoreduction efficiency than undoped TiO 2under UV–vis irradiation.I-TiO 2calcined at 375◦C has superior activity to those calcined at higher temperatures.Optimal doping levels of iodine were identi-fied under visible and UV–vis irradiations,respectively.This is the first study that investigates nonmetal doped TiO 2without other co-catalysts for CO 2photoreduction to fuels under visible light.© 2011 Elsevier B.V. All rights reserved.1.IntroductionThe global warming effect is believed to be associated with the increasing concentrations of greenhouse gases in the atmo-sphere,where the major contribution comes from CO 2emissions from fossil fuel consumption.Current pre-and post-combustion CO 2capture and sequestration technologies are energy intensive and thus costly [1].Furthermore,there are many uncertainties with regard to long-term storage of CO 2in geological formations [1].In contrast,with recent innovations in photocatalysis,recy-cling CO 2to fuels using sunlight as the sole energy input offers a brand new opportunity for a sustainable energy future.This can not only mitigate CO 2emissions but also produce energy-bearing compounds such as CO,methane,and methanol [2–4]that can be subsequently converted to liquid transportation fuels.Materi-als that have been reported for CO 2photoreduction applications include ZrO 2[5,6],MgO [7],NiO/InTaO 4[8],Ga 2O 3[9],photosen-sitized complexes [10],and TiO 2based catalysts [11–13].TiO 2is a∗Corresponding author.Tel.:+14142293716;fax:+14142296958.E-mail address:liying@ (Y.Li).promising candidate because of its strong redox ability,low cost,stability,and environmental benignness.However,one challenge for the application of semiconductor photocatalysts like TiO 2is the fast recombination of photo-induced holes (h +)and electrons (e −).Another challenge is the requirement of ultraviolet (UV)light exci-tation due to the wide band gap of TiO 2(3.2eV for anatase and 3.0eV for rutile).As a result,the efficiency of CO 2conversion to fuels is generally low.Modification of TiO 2with metal (e.g.Pt,Pd,Ag,and Cu)parti-cles or clusters has been reported to inhibit charge recombination possibly because the metals serve as electron traps [14].Thus,an increased CO 2photoreduction efficiency was observed for metal modified TiO 2.Tseng et al.[3]synthesized Cu/TiO 2catalysts by a sol–gel method and found the rate of CO 2photoreduction to methanol was much higher than those without copper loading.Li et al.[13]reported markedly increased CO 2photo-conversion efficiency by Cu/TiO 2catalyst dispersed on mesoporous silica and selective CH 4production due to Cu loading.However,too high a concentration of metal dopant may form recombination centers that lead to a reduced photocatalytic efficiency.Opti-mal metal concentrations have been reported for modified TiO 2(e.g.with Ag or Cu)for both photooxidation and photoreduction0926-860X/$–see front matter © 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.apcata.2011.04.032196Q.Zhang et al./Applied Catalysis A:General400 (2011) 195–202applications[13,15–17].While metal modifications on TiO2have apparent enhancement in charge separation,they have limited con-tribution to extending the photo-response to visible light region. Sasirekha et al.[18]observed that Ru doped TiO2has almost the same absorption spectra as the undoped TiO2.Dholam and Patel [19]reported that Cr and Fe doped TiO2prepared by a sol–gel method had a very limited effect on inducing a red-shift in TiO2 absorption spectra compared to those prepared by a magnetron sputtering method.On the other hand,it has been widely reported that doping or co-doping TiO2with nonmetals(e.g.C,N,S,F,etc.)has resulted in more significant band gap narrowing compared to metal doping, leading to high photocatalytic efficiency under visible light irra-diation[20–27].Wu et al.[26]reported that the band gaps of N doped and N–B co-doped TiO2were2.16eV and2.13eV,respec-tively,much smaller than that of pure TiO2(3.18eV for anatase). Pelaez et al.[27]synthesized N–F co-doped TiO2that exhibited high surface area,low degree of agglomeration and high activity in degradation of microcystin under visible light.Recently,less studied iodine has been doped into TiO2,and improved visible light activity towards the decomposition of organic compounds has been reported[28–30].The structural and electronic proper-ties of I-doped TiO2were investigated based on density functional theory(DFT)calculations,and the results indicated that substi-tutional iodine contributes to a much more efficient and stable photocatalyst than pristine TiO2[31].In comparison to other non-metal dopants(N,C,B,S),iodine doping may result in superior photocatalytic activity due to the following reasons.First,unlike other nonmetal dopants that substitute lattice oxygen,iodine was reported to be able to replace lattice titanium due to the close ionic radii of I5+and Ti4+[28,32].The substitution of Ti4+with I5+ causes charge imbalance and results in the generation of Ti3+sur-face states that may trap the photoinduced electrons and forestall charge recombination[32].In addition,first principle calculations suggest that iodine atoms prefer to be doped near the TiO2surface due to the strong I–O repulsion[29],and thus,the surface doped I5+will not only trap electrons but also facilitate electron transfer to the surface adsorbed species[29,33].Finally,it is suggested that the continuous states consisting of5p and/or5s orbitals of I5+and O2p orbitals of the valence band are favorable for efficient trap-ping of holes at the I-induced states in the TiO2particle(not on the surface),which causes a decrease in the oxidation power[33].For CO2photoreduction,one of the challenges is the re-oxidation of the CO2reduction products by h+or OH•radicals.Hence,the impaired oxidation power of I-doped TiO2may result in an increased CO2 photoreduction rate.In this study,I-doped TiO2photocatalysts were synthesized via a hydrothermal method and were evaluated for thefirst time for CO2reduction under UV and visible light irradiation.The effects of iodine doping levels and calcination temperature on the cat-alytic activity were also investigated,a topic that has been scarcely discussed in the literature for I-doped TiO2.The structural and com-positional properties of the I-TiO2nanomaterials were analyzed and correlated with their photocatalytic reduction performance. This is thefirst paper that reports photocatalytic CO2reduction by nonmetal doped TiO2without any other co-catalysts.There-fore,thefindings are an important step towards the discovery of cost-effective catalysts for CO2reduction to solar fuels.2.Experimental2.1.Photocatalyst preparationThe method for synthesis of I-doped TiO2was modified from that reported by Tojo et al.[33].The preparation started by dissolv-ing3ml of titanium isopropoxide(TTIP)(Acros Organic,>98%)in 3ml of anhydrous isopropanol(Acros Organic,>99.5%).The mix-ture was then added dropwise into a solution of iodic acid(HIO3) (Alfa Aesar,>99.5%)with continuous stirring for2h.After the reac-tion,the resultant white mixture was transferred to a Teflon-lined vessel for hydrothermal treatment at100◦C for12h.The resultant yellow particles werefiltrated and washed with copious amount of de-ionized water until pH7followed by drying in an oven at80◦C for1h.The samples werefinally calcined in air for2h at different temperatures(375,450,or550◦C).Different iodine doping levels were prepared by varying the quantity of HIO3added(0.06–0.18g). The samples are denoted in the way of“x%I-TiO2-y C”,where x is the nominal weight percentage of iodine in the sample(calculated from the bulk solution)and y is the calcination temperature.For example,5%I-TiO2-375C represents5wt.%(nominal)I-doped TiO2 calcined at375◦C.For comparison,undoped TiO2was prepared following the same procedure without adding HIO3.All samples were grinded and sieved by a45␮m stainless steel sieve before characterization and photoreduction experiments.2.2.Photocatalyst characterizationBrunauer–Emmett–Teller(BET)surface area analysis by N2 adsorption was performed using a Quantachrome NOVA1200gas sorption analyzer(Boynton Beach,FL).The crystal structures of the prepared catalysts were identified by X-ray diffraction(XRD)(Scin-tag XDS2000)using Cu K␣irradiation at45kV and a diffracted beam monochromator operated at40mA in the2Ârange from20◦to70◦at a scan rate of1◦/min.The crystal size of different crys-tal phases was calculated by the Scherrer equation.The fractional phase content,W A,W B,and W R,for anatase,brookite,and rutile, respectively,are mathematically defined in Eqs.(1)–(3)[34]:W A=0.886×A A(0.886×A A+A R+2.721×A B)(1)W B=2.721×A B(0.886×A A+A R+2.721×A B)(2) W R=A R(0.886×A A+A R+2.721×A B)(3)where A A,A B and A R represent the integrated intensity of the anatase(101)peak(2Â=25.28◦),the brookite(121)peak (2Â=30.81◦),and the rutile(110)peak(2Â=27.45◦),respec-tively.Because the brookite(120)(2Â=25.34◦)and brookite(111) (2Â=25.69◦)peaks overlap with the anatase(101)peak,A A and A B were calculated by the following ing the single isolated brookite(121)peak as a reference,the anatase(101), brookite(120)and brookite(111)overlapped peaks were decon-voluted by0.9and0.8intensity ratio for I(121)(brookite)/I(120)(brookite)andI(111)(brookite)/I(120)(brookite)respectively,with the same FWHM of brookite (121)[35].The lattice structure of individual nanocrystals was visualized by phase-contrast high resolution transmission electron microscopy (HRTEM)carried out with300keV electrons in a Hitachi H9000NAR instrument with0.18nm point and0.11nm lattice resolution. Two-dimensional Fourier transforms were calculated and used to measure lattice spacing and interplanar angles.Amplitude con-trast TEM images were used to obtain direct information about the nanocrystal sizes.Selected area electron diffraction(SAD)pro-vided information that is analogous to XRD,but from nanocrystals supported on an electron-transparent amorphous carbonfilm and selected within a∼450nm diameter aperture.The UV–vis diffuse reflectance spectra were obtained by a UV–vis spectrometer(Ocean Optics)using BaSO4as the back-Q.Zhang et al./Applied Catalysis A:General 400 (2011) 195–202197Fig.1.Experimental setup for CO 2photoreduction.1:Mass flow controller;2:water bubbler;3:photoreactor with a quartz window;4:two-way valve;5:long-pass filter;6:gas chromatograph (GC/TCD-FID);7:catalyst samples dispersed on glass-fiber filter;8:Xe lamp;9:sampling port.ground.The reflectance was converted to F (R )values using the Kubelka–Munk function (Eq.(4)).F (R )=(1−R )22R =k s(4)where R is the absolute reflectance,k is the molar absorptioncoefficient and s is the scattering coefficient.The band gaps were obtained from the plot of [F (R )E ph ]1/2against the photon energy E ph .X-ray photoelectron spectroscopy (XPS)analysis was carried out on a PerkinElmer PHI 5100ESCA system with an Al K ␣X-ray source (h =1253.6eV)and pass energy of 35.75eV operating at a pressure of 8×10−10Torr.The observed spectra were corrected with the C1s binding energy (BE)value of 284.6eV.2.3.CO 2photoreduction experimentsThe schematic of the photocatalytic reaction system is illus-trated in pressed CO 2(99.99%,Praxair)regulated by a mass flow controller was passed through a water bubbler to gener-ate CO 2and H 2O vapor mixture (H 2O,v/v%≈2.3%).The gas mixture was then purged through a cylindrical photoreactor (V =58cm 3)with stainless steel walls and a quartz window.A fixed amount of powder catalyst (200mg)was dispersed on a glass-fiber filter and placed at the bottom of the reactor.After purging for 1.5h,the gas valves on both sides of the reactor were closed to seal the reactor.A 450W Xe lamp (Oriel)was used as the light source and a long-pass filter was applied to cut off the short wavelengths that are less than 400nm if only visible light is needed.A spectroradiome-ter (International Light Technologies ILT950)was used to obtain the spectral intensity of the Xe lamp with and without the filter.During the illumination period,the gaseous samples in the reactor were taken by a gastight syringe (Hamilton,#1750,500␮l)every 30min and manually injected to a gas chromatograph (GC,Agilent 7890A)equipped with both a thermal conductivity detector (TCD)and flame ionization detector (FID).Prior to CO 2photoreduction experiments,all catalyst powders were pre-treated under UV irradiation (12W,365nm)for 12h to eliminate organic residues on the catalyst,if any.GC measurements were also performed using a mixture of ultra high purity helium (instead of CO 2)and water vapor as the purging and reaction gas for the catalyst-loaded reactor;no carbon-containing compounds were produced by the catalyst under UV or visible irradiation.This verifies that the catalyst was clean (i.e.no interference from organic residues).A series of other background tests were also conducted using a mixture of CO 2and H 2O vapor as the purging and reaction gas for both cases of (1)empty reactor and (2)blankglass-fiberFig.2.XRD patterns of I-doped TiO 2at different calcination temperatures (a)and I-doped TiO 2at different iodine doping level (b)(A:anatase;B:brookite;R:rutile).filter in the reactor.Again,for either case no carbon-containing compounds were produced under either UV or visible irradiation.This demonstrates that the reactor and the glass-fiber filter were clean and that the CO 2conversion cannot proceed without the photocatalyst.All these background tests have proved that any carbon-containing compounds produced must be originated from CO 2through photocatalytic reactions.3.Results and discussion3.1.Average nanocrystal structure from XRD analysisFig.2shows the XRD patterns of TiO 2samples doped with dif-ferent concentrations of iodine calcined at 375◦C and 5%I-TiO 2calcined at different temperatures.The calculated values of phase content and crystal size are listed in Table 1.The undoped and I-doped TiO 2mainly consist of two phases,anatase and brookite.As the calcination temperature increased from 375to 450◦C,the anatase phase content increased and the brookite phase decreased;when the calcination temperature increased to 550◦C,the brookite content further decreased with the appearance of a small percent-age of rutile.The phase transition between metastable anatase and brookite is not well studied in the literature;however,the result in this study seems to be in agreement with some of the litera-ture that upon calcination brookite transforms to rutile via anatase [36–38].In other words,brookite first transforms to anatase and then anatase transforms to rutile.In contrast to undoped nanocrystal TiO 2whose anatase-to-rutile transformation temper-ature is around 700◦C [39],the lower temperature (450–550◦C)of transformation to rutile in this study was possibly due to the dopant-induced instability of TiO 2caused by lattice distortion and bond weakening [33],even at a low dopant concentration (for 5%I-TiO 2).When the calcination temperature was kept the same at 375◦C,increasing the iodine concentration only slightly decreased the brookite content by a few percent.198Q.Zhang et al./Applied Catalysis A:General400 (2011) 195–202Table1Phase content and average crystal size of I-TiO2samples obtained from X-ray diffraction,band gap from optical spectroscopy,and specific surface area from BET analysis (A:anatase,B:brookite,R:rutile).Sample Phase content(%)Crystal size(nm)Band gap(eV)BET specific surface area(m2/g)A B R A BTiO2-375C663408.8 4.7 3.13122.95%I-TiO2-375C663407.5 5.4 3.05137.45%I-TiO2-450C712908.410.1–99.45%I-TiO2-550C7619519.112.7–43.110%I-TiO2-375C70300 5.5 6.3 3.00137.615%I-TiO2-375C72280 5.8 5.5 3.02137.6In terms of crystal size,it is clear from Table1that with increas-ing calcination temperature the average crystal size increases for both anatase and brookite crystals(by a factor of∼2.45±0.10at 550◦C compared with375◦C).At the same calcination tempera-ture of375◦C,the undoped TiO2has the largest crystal size for anatase(8.8nm),and the crystal size decreases(to7.5,5.5,and 5.8nm)as the nominal iodine doping level increases in increments of5%(from0%to15%).This result agrees with the literature that dopants can favor the formation of smaller particles[40].For exam-ple,Zhou et al.[22]reported the particle sizes of N doped and N–I co-doped TiO2are smaller than pure TiO2.Su et al.[30]found that I-doped TiO2has much smaller crystallite size(7.7nm,anatase)than undoped TiO2(23.7nm,anatase)and suggested that the repulsion among adsorbed iodine species inhibits crystal growth.An interest-ingfinding in our study is that the brookite crystallite size of I-doped TiO2is slightly larger than that of undoped one(Table1),indicat-ing that iodine has opposite effect on the growth rate of anatase and brookite nanocrystals under otherwise identical hydrother-mal conditions.The lack of literature on doping of brookite TiO2 warrants further investigation in this interesting phenomenon. 3.2.Individual nanocrystal structure and morphology from TEM analysisFig.3shows amplitude-contrast transmission electron microscopy(TEM in(a))and phase-contrast high-resolution TEM images(HRTEM in(b)and(c))of the5%I-TiO2-375C sample. Both types of images show agglomerates of TiO2nanocrystals. The crystallite size is in the range of6–9nm,which is in good agreement with the average size calculated from the Scherrer equation.Similarly,selected area electron diffraction experiments (SAED inset in(a))recorded from agglomerates within a selecting aperture of450nm confirm the phase determination of XRD and demonstrates that the anatase and brookite phases of I-TiO2 occur in close proximity.The HRTEM images show lattice fringes within individual nanocrystals.Analysis of the lattice spacings and interpanar anglesfinds that each nanocrystal has a well-defined phase and the lattice appears cleanly and bulk-terminated at the surface.The nanocrystal morphology is defined by low-energy facets that are conjoined by curved surfaces composed of closely spaced terraces and steps.For example,the HRTEM image in Fig.3c shows clear one-dimensional lattice fringes of TiO2(lattice spac-ing=0.345nm)which is very close to the brookite TiO2(111)bulk lattice spacing of0.346nm,according to powder diffractionfile (PDF)No.29-1360.Since all of the anatase spacings overlap with brookite very closely,it is only possible to uniquely determine the termination facets of the brookite nanocrystals.These consistently yield the(111)type of crystal plane as dominant facet for the brookite TiO2nanocrystals.The second type of termination plane occurs for interplanar distance of∼0.351nm which is the(101) plane of anatase or the(120)plane of brookite.It is unlikely that brookite nanocrystals would have two very different dominant facets under the same growth and calcination condition.Hence,Fig.3.Electron microscopy of5%I-TiO2-375C sample:(a)TEM image and SAED (inset),(b)HRTEM image with labeled examples of anatase(A)and brookite(B) nanocrystals,and(c)HRTEM lattice spacings and dominant surface facets for A(120) and B(111)nanocrystals,with arrows pointing at steps on B(111)surface.Q.Zhang et al./Applied Catalysis A:General400 (2011) 195–202199Fig.4.UV–vis diffuse reflectance spectra of TiO2with different iodine doping levels (a)and plots of the square root of the Kubelka–Munk function versus the photon energy(b).it is possible to conclude,by elimination,that the second type of facets belong to anatase(101)type planes.These morphology changes are the subject of on-going work and are beyond the scope of this initial work to evaluate the efficacy of I-doped TiO2as an effective material for CO2photoreduction.3.3.Band gap analysis from UV–vis diffuse reflectance spectroscopyThe UV–vis diffuse reflectance spectra of undoped TiO2and I-doped TiO2samples are shown in Fig.4a.The absorption edge of undoped TiO2is around400nm and is extended to the visible light region for I-doped TiO2with an iodine concentration from2.5to 15%,which matches the yellow color of the I-doped TiO2.Fig.4b illustrates the plots for obtaining the band gap values that are also listed in Table1.The undoped TiO2has a band gap of3.13eV,while the band gaps of I-TiO2slightly decrease with iodine doping and level off at around3.00eV when the nominal iodine concentration is greater than10%.3.4.BET specific surface areaThe BET specific surface areas(SSA)of the various I-TiO2cata-lysts are listed in Table1.The SSA for undoped TiO2is122.9m2/g, which is much higher than that of commercially available P25 (∼50m2/g).With iodine doping in TiO2,the SSA slightly increases as the crystal size slightly decreases in average,and all the I-TiO2 samples calcined at the same temperature(375◦C)have similar SSA (∼137m2/g)since their average crystal sizes are very close to each other,as seen in Table1.Increasing the calcination temperature of 5%I-TiO2to450and550◦C dramatically reduces the SSA to99.4and43.1m2/g,respectively,which corresponds well to the increase in crystal size.3.5.Host and dopant valence from XPS studiesFig.5a shows XPS survey spectrum of10%I-TiO2calcined at375◦C which indicates the existence of Ti,O,and I ele-ments.Fig.5b–d shows the high resolution XPS spectra scanning over the following three binding energy areas:(1)Ti2p region (450–470eV);(2)I3d region(610–640eV);and(3)O1s region (520–540eV).Table2summarizes the surface atomic concentra-tions of Ti,I,and O as well as the atomic percentages of the three elements at different chemical states.As shown in Fig.5b,the XPS peaks of the10%I-TiO2-375C sam-ple in the Ti2p region appear at458.5eV(Ti2p3/2)and464.2eV(Ti 2p1/2),both of which correspond to Ti4+.There are two small peaks in the lower side of Ti2p3/2(457.5eV)and Ti2p1/2(463.0eV),which are ascribed to Ti3+that is generated to maintain the electroneutral-ity by I5+substituting Ti4+,as the ionic radii of I5+and Ti4+are very close[32,41].The XPS spectra of I3d region(Fig.5c)for10%I-TiO2-375C show double peaks around623.6eV(I3d5/2)and635.1eV(I 3d3/2)which infer that the oxidation state of doped iodine is I5+ [32,33].Also two weaker satellite peaks around619.9eV(I3d5/2) and631.4eV(I3d3/2)indicate existence of I−[33,41].The results agree with the literature that I5+/I−pairs were observed for I-doped TiO2and that I5+ions substitute for Ti4+and are present in the I–O–Ti bond[28,33,41].Some other studies have reported TiO2 doped with multivalency iodine I7+/I−prepared by different meth-ods or using different precursors[22,30].The I7+peak at around 624.0eV(I3d5/2)was not observed in our study.The XPS spectra of the O1s region(Fig.5d)show a major peak at around529.5eV that that correspond to lattice oxygen O2−.The other peak around 530.8eV can be attributed to chemisorbed oxygen on the surface [22]or surface hydroxyl groups[33].From Table2,it is clear that the surface atomic concentration of iodine increased with the nominal iodine doping level from0%to 15%.The I/Ti atomic ratio on the surface is much larger than that in the bulk(nominal)for the10%and15%I-TiO2samples,indicat-ing that iodine is mainly doped on the surface.The percentage of Ti3+/Ti increased to9.9%,18.4%and19.8%as the nominal iodine con-centration increased to5%,10%and15%by weight.This supports our conclusion that I5+substitutes Ti4+and results in generation of Ti3+due to charge imbalance.For the15%I-TiO2sample,the sur-face iodine concentration(2.5at.%)and Ti3+fraction(19.8at.%)are only slightly larger than those of the10%I-TiO2sample(2.3at.% and18.4at.%,respectively),suggesting that the doping of iodine approaches to a saturation level for the15%I-TiO2sample.This also correlates with the result that the band gap of I-TiO2decreases with increased iodine concentration but levels off at the concentration of iodine greater than10wt.%(Table1).3.6.Photocatalytic activity for CO2reductionIn this study,CO was identified as the main CO2reduction prod-uct using undoped and I-doped TiO2,while our previous study [13]showed that CH4was produced in addition to CO when the TiO2surface was loaded with Cu species.The following reactions may express the pathways of CO2photoreduction to CO and water oxidation to O2:TiO2hv−→e−cb+h+vb(R1) 2H2O+4h+→4H++O2(R2) CO2+2H++2e−→CO+H2O(R3)200Q.Zhang et al./Applied Catalysis A:General400 (2011) 195–202Fig.5.XPS spectra of 10%I-TiO 2calcined at 375◦C:survey spectrum (a)and high-resolution spectra for Ti 2p (b),I 3d (c),and O 1s (d).Fig.6shows the concentration of CO produced in the reactor (in ppm)as a function of illumination time under visible ( >400nm)and UV–vis irradiation ( >250nm),respectively.Undoped TiO 2had no activity under visible light (results not shown in Fig.6a).All I-doped TiO 2showed visible light activity for CO 2photoreduction to CO and the concentration of CO increased almost linearly with illumination time (Fig.6a).With the same iodine concentration (5%),I-TiO 2calcined at 375◦C had the highest activity;increased calcination temperatures (450◦C and 550◦C)lowered the CO 2pho-toreduction rate.This is likely due to the significant increase of the crystal size and decrease in surface area with increasing calcination temperature (Table 1).For I-TiO 2with the same calcination tem-perature (375◦C),activity of CO 2photoreduction follows the order of 10%>15%>5%in terms of iodine doping concentration (Fig.6a).The band gap analysis (Table 1)and XPS analysis data (Table 2)show that 10%and 15%I-TiO 2samples have very close band gap energies (3.00eV and 3.02eV,respectively)and surface iodine concentra-tions (2.3at.%and 2.5at.%,respectively),suggesting that the 15%I-TiO 2sample may not be superior to 10%I-TiO 2.Furthermore,too high a dopant concentration may form charge recombination cen-ters and/or shield the surface of TiO 2from light irradiation,both of which reduce the photocatalytic activity.These may explain why 10%corresponds to the optimal iodine concentration under visible light irradiation.Similar findings on the optimal dopant concen-tration have been reported for metal-doped TiO 2[3,13,17,41,42],while optimal doping concentrations of nonmetals have been muchless discussed possibly because they are more difficult to control.This is the first study that has reported an optimal doping level for I-doped TiO 2.The concentration of CO reached 670ppm at 210min for the sample of 10%I-TiO 2-375C under visible light (Fig.6a),resulting in a product yield equivalent to 2.4␮mol g −1h −1.There have been very few studies on the CO 2photoreduction under visible light irradiation.This paper,for the first time in the literature,reports photocatalytic CO 2reduction by nonmetal-doped TiO 2without any other co-catalysts.Grimes and co-workers [43]synthesized N-doped TiO 2nanotube arrays sputtered with Cu nanoparticles (NT/Cu)for CO 2photoreduction under sunlight and they reported that the activity in visible light region is only 3%of that under the whole solar spectrum (the rates of CO,CH 4,and other HCs production are equivalent to 0.11,0.13,and 0.05␮mol g −1h −1,respectively;the total production rate is about 0.3␮mol g −1h −1under visible light irradiation of the solar spectrum with an inten-sity of 78.5mW/cm 2)[43].Ozcan et al.[44]studied dye-sensitized and Pt modified TiO 2for CO 2photoreduction and reported a CH 4production rate of 0.2␮mol g −1h −1using a 75W daylight lamp as the visible light source.Fig.7shows the spectra of the 450W Xe lamp used in this work,with or without the 400nm long-pass fil-ter,in comparison with the AM 1.5G standard solar spectrum.The integrated light intensity of the Xe lamp was 428mW/cm 2(full spectrum)and 233mW/cm 2for the visible region (400–750nm).While the visible light intensity in our study was approximatelyTable 2Surface atomic concentration of I-TiO 2catalysts from XPS analysis.SampleSurface atomic concentration (%)I/Ti atomic ratio (%)Ti species (%)I species (%)O species (%)ITi O Nominal Surface Ti 4+Ti 3+I 5+I −O lattice O absorb TiO 2-375C017.182.9––1000––65.334.75%I-TiO 2-375C 0.417.482.2 3.3 2.390.19.916.084.057.642.410%I-TiO 2-375C 2.320.277.57.011.481.618.469.930.156.343.715%I-TiO 2-375C2.515.981.611.115.780.219.857.342.762.237.8。