表面缺陷與鈀的沉積對硫化鎘納米晶粒的光催化制氫性能的影響

劉志明，劉國亮，洪昕林

武漢大學(xué)化學(xué)與分子科學(xué)學(xué)院，武漢 430072

1 Introduction

The development of renewable energies has attracted immense attention in recent years.Among them,the photocatalytic hydrogen production from water splitting via converting light into chemical energy is a very potential process to meet future energy requirement1–5.Since TiO2had been first used as a catalyst in photocatalytic water splitting to produce hydrogen,numerous semiconducting catalysts have been developed in the past few decades6–10.As we know,the visible light region account for 44% of the total sunlight irradiation,much larger than ultraviolent wavelength region (3%).In terms of a high utilization efficiency of solar energy,the development of visible light responsive photocatalysts is of great importance11–18.

Cadmium sulfide (CdS) possesses a narrow band gap (2.4 eV)and it can absorb visible light lower than 520 nm19,20.The visible light response and proper conduction band energy level (above the H+/H2potential) ensure CdS to be a good candidate for photocatalytic hydrogen production from water splitting19–24.Although CdS is a good photocatalyst,it still shows limited activity in water splitting to produce hydrogen,due to the fast recombination of photo-induced electron/hole pairs and the low hydrogen evolution reaction rate on the surface.To improve quantum yield in photocatalysis,the surface of a catalyst normally requires enough “charge” traps to decrease the recombination probability of photo-induced electrons/holes or highly reactive sites to consume the charges for redox reactions25.To better understand the relationship between catalyst structure and activity,the fabrication of a proper surface structure model is of great importance,as the solid surface is very complicated,which contains surface defects,heterojunction structure,crystal facet,and etc.It is generally accepted that surface defects can trap photo-induced electrons or holes26,27,thus facilitating their separation on catalyst surface for further redox reaction.Despite great progresses have been achieved in recent years,the studies on the effect of surface atomic structure and properties are still very limited.

In this work we compared three CdS nanocrystals with different morphologies including long rod,short rod and triangular plate (denoted as lr-CdS,sr-CdS and tp-CdS,respectively) in the photocatalytic hydrogen production reaction.The exposed surface (polar versus nonpolar surface) varies with the aspect ratio of CdS.UV-Vis adsorption and photoluminescence spectra provide information on the band structure and surface defects.It is found that the degree of surface defects is associated with the hydrogen production rate.Surface defects can trap photo-induced electrons/holes,and thus decrease their probability of recombination,which may explain the high activity.Moreover,taking the advantage of the surface defects,we obtained the Pd deposition onto CdS surface.This greatly enhanced the hydrogen production rate,with highest value reached 7884 μmol·h-1·g-1over sr-CdS/Pd,43 times higher than that of pure sr-CdS.Such excellent activity may be explained by the heterojunction structure.

2 Experimental

2.1 Materials and sample preparation

All the reagents were of analytical grade from Sigma-Aldrich,and were used without further purification.

Triangular plate-like CdS (tp-CdS) was synthesized by using a modified solvothermal method from reference28.In brief,40 mL oleylamine (98%),2 mmol Cd(OAc)2·2H2O (98.0%),6 mmol sulfur powder (99.9%),and 6 mmol triphenylphosphine(99%) were added to 30 mL toluene (99.8%).After 10 min of stirring,the solution turned clear and was transferred into a Teflon-lined stainless steel autoclave with a capacity of 125 mL.The autoclave was sealed,heated,and maintained at 190 °C for 24 h.Then it was allowed to cool to room temperature naturally.The resulting bottom precipitates were washed with toluene several times,and then dispersed in hexane followed by centrifugation to remove undissolved precipitate.A proper amount of ethanol was added into the dispersion followed by centrifugation at 5000 r·min-1.The resulting sample was dried in a vacuum container for 12 h.

Long rod-like CdS (lr-CdS) was prepared by using a modified solvothermal method from reference29.5 mmol Cd(OAc)2·2H2O and 7.5 mmol sulfur powder added dispersed in 25 mL liquid dodecylamine (99%) at 50 °C.After 10 min of stirring,the solution turned homogeneous and was transferred into a Teflonlined stainless steel autoclave with a capacity of 45 mL.The autoclave was sealed,heated,and maintained at 100 °C for 10 h.Then it was allowed to cool to room temperature naturally.The resulting bottom precipitates were washed with ethanol several times.The resulting sample was dried in a vacuum container for 12 h.

Short rod-like CdS (sr-CdS) was prepared by using a modified solvothermal method from reference29.18 mmol Cd(OAc)2·2H2O and 12 mmol sulfur powder added dispersed in 60 mL liquid dodecylamine at 50 °C.After 10 min of stirring,the solution turned homogeneous and was transferred into a Teflon-lined stainless steel autoclave with a capacity of 125 mL.The autoclave was sealed,heated,and maintained at 220 °C for 10 h.Then it was allowed to cool to room temperature naturally.The resulting bottom precipitates were washed with ethanol several times.The resulting sample was dried in a vacuum container for 12 h.

PdS/CdS nanocomposites were realized by in-situ deposition of 1% (w,mass fraction) PdS on CdS,just prior to the photocatalytic reaction.Typically,a Pd(NO3)2 aqueous solution was added dropwise to a suspension of CdS powder (20 mg,as prepared above) dispersed in an aqueous solution consisting 0.35 mol·L-1Na2S and 0.25 mol·L-1Na2SO3under sonication30,31.

2.2 Photocatalytic test

Photocatalytic reactions were performed in a 250 mL roundbottomed flask by dispersing 20 mg of catalysts in a 100 mL aqueous solution containing 0.35 mol·L-1Na2S and 0.25 mol·L-1Na2SO3as sacrificial agents.The solution was bubbled with 5%CH4/Ar mixed gas for 30 min to remove the dissolved oxygen prior to irradiation under a 300 W white light source (λ varies from 400 to 800 nm).The distance between reactor and white light source was fixed at 20 cm.During the irradiation process,agitation of the solution ensured uniform irradiation of the samples.

After a reaction time of 2 h,a 30 mL of the generated gas sample was collected and analysed by gas chromatograph(Agilent 7890A).The composition of gas produced from all samples was tested using a Gas Chromatograph (GC) fitted with both a Thermal Conductivity Detector (TCD) and a Flame Ionisation Detector (FID).This allowed for accurate detection of H2,Ar,CH4and identification of other species that may have been present (N2,CO2 and O2).

2.3 Characterizations

Transmission Electron Microscopy (TEM):A JEOL 2010 was used to record TEM images.The samples were sonicated for 1 h in ethanol and dropped onto a lacey carbon coated copper grid,then dried in air at room temperature.

Powder X-Ray Diffraction (XRD):XRD analysis was performed using a PANalytical X’Pert Pro diffractormeter,operating in Bragg-Brentano focusing geometry and using Cuka radiation (λ = 0.15418 nm) from a generator operating at 40 kV and 40 mA.The sample powder was placed directly onto a glass slide and smoothed to form a thin uniform layer.

X-Ray Photoelectron Spectroscopy (XPS):XPS was performed in a VG Microtec ion pumped XPS system equipped with a nine channel CLAM4 electron energy analyser.200 W Mg X-ray excitation was used.The samples were analyzed with reference to adventitious carbon 1s peak.

Photoluminescence (PL) spectra:PL spectra were carried out using a Fluo Time 300 spectrum instrument.The samples were coated on a transparent glass slice before tests.The excitation wavelength was set at 405 nm.

UV-Vis diffuse reflectance spectra (UV-Vis DRS):UV-Vis DRS tests were carried out on a LAMBDA35 spectrum instrument (Perkin Elmer).The scanning wavelength ranges from 250 to 800 nm and the resolution was set at 1 nm.

Fig.1 XRD patterns of lr-CdS,sr-CdS and tp-CdS.

3 Results and discussion

3.1 Material characterization

XRD Characterization was first used to examine the structure of all synthesized CdS samples,as shown in Fig.1.For all samples,wurtzite phase structure of CdS (JCPDS#02-0549) can be determined by diffractions at 2θ = 24.9°,26.7°,28.3°,36.8°,43.9°,48.1°,52.1° from XRD patterns.For the tp-CdS nanocrystals,the weak and broad peaks suggest a low crystallinity of the CdS sample.The diffraction peaks become more intense for the two rod-like samples (lr-CdS and sr-CdS),indicating their nanostructures get higher crystallinity.By comparing the relative strength and full width at half maximum(FWHM) of the (002) peak,we can obtain the crystal stacking information along c-axis direction [001].For tp-CdS,the (002)peak is stronger and broader than the (100) and (101) peaks,indicating that the sample has smaller grain size along the c-axis direction.As the c-direction size increases,the (100) and (101)peaks intensifies,indicating an increase of the diffraction orientations along (101) and (100) directions in the sample.For lr-CdS,the FWHM of 002 diffraction is much narrower than(100) and (101),indicating a larger crystal size along c-axis with a long rod structure.

Fig.2 shows conventional TEM images and corresponding high-resolution images of three CdS samples.Fig.2a clearly shows rather regular triangular plate-like structures,with an average lateral size of 18 nm.From the high resolution image in Fig.2b,we can clearly see a hexagonal pattern of surface atoms and the same d-spacing value of 0.36 nm of stacked fringes along three directions.This matches well with the characteristic (100)facets of wurtzite structure of CdS,indicating a top view through c-axis direction.This can be further confirmed by the corresponding fast Fourier transform in the inset of figure b,revealing a hexagonally symmetric pattern of electron diffraction.So,(002) facet is the major exposed surface for tp-CdS28.Fig.2c displays an image of sr-CdS sample.Most of the crystals show short rod structures with narrow size distribution,of about 18 nm in diameter and 40 nm in length,and of an aspect ratio of around 2.2.Interestingly,the rods tend to form multipodlike structures.Each branch shows a lattice d-spacing of 0.336 nm (in Fig.2d),suggesting a crystal growth along the [001]direction (c direction).Fig.2e shows a TEM image of lr-CdS.The crystals follow a similar growth mechanism as that of sr-CdS,confirmed by the measured lattice fringe of 0.336 nm (Fig.2f).At a lower temperature,long rods are formed,of 5 nm in diameter and 50 nm in length,with an aspect ratio of 10.0.The lr-CdS mainly expose (100) or (110) facet.In literature29,the formation of the nanocrystals in the reaction solution is usually divided into nucleation and growth processes.Both multipodlike samples are formed by starting from triangular nuclei which are shape-determinant,followed by surface-initiated growth of arms.A lower temperature would favor a kinetic-control growth process for anisotropic structure,thus contributing to a higher aspect ratio.

To evaluate the chemical states of the CdS,XPS analyses were conducted over all synthesized CdS samples,as shown in Fig.3.The peaks at 405.0 and 411.8 eV are attributed to Cd2+,while those at 161.5 eV are attributed to S2-.There are no apparent differences in the binding energy in terms of both Cd and S for the three CdS samples,indicating that all of them were successfully synthesized.

Fig.4a shows the diffuse reflectance spectra (DRS) of three CdS samples.For all samples,the absorption edge is located at around 510 nm,suggesting that they can absorb visible light with λ ＜ 510 nm.Using the Kubelka-Munk equation32,the bandgap is calculated to be 2.51,2.50 and 2.54 eV for tp-CdS,sr-CdS and lr-CdS,respectively.Among them,lr-CdS shows a slightly wider bandgap due to quantum confinement caused by its higher aspect ratio.

Fig.2 TEM (up) and the corresponding HR-TEM (down) images of (a,b) tp-CdS,(c,d) sr-CdS and (e,f) lr-CdS.Inset of the figure b is the fast Fourier transform pattern.

Fig.5 shows the room-temperature photoluminescence (PL)spectra of CdS with different shape.For tp-CdS,the curve shows two distinct regions,a narrow one in the short-wavelength range and a broadened one in the long-wavelength range,similar to the previous reports of triangular plate-like CdS28.The narrow peak at around 520 nm is assigned to near-band-edge emission (NBE),while the broad one (600–900) can be associated with the structural defects emission (DE) which may arise from the excess of sulfur or core defects on the CdS surfaces29.Obviously,the intensity of NBE is much lower than DE,making their intensity ratio (INBE/IDE) equal to 0.4,suggesting that the material possessing a lot of defects on the surface.The sr-CdS sample shows a relatively weak NBE signal.The INBE/IDEratio decreases to 0.2,revealing a higher concentration of surface defects than that of tp-CdS.When the aspect ratio becomes higher,the lr-CdS has a negligible NBE signal which exhibits an obvious blue-shift to around 470 nm.Meanwhile,the defect emission also shifts towards short wavelength region,quite in agreement with reported CdS nanowires.The blue shift of PL can be explained by the quantum confinement.By comparing the INBE/IDEratio between three samples,we can obtain an order of surface defects,lr-CdS ＞ sr-CdS ＞ tp-CdS.

Fig.3 XPS spectra of (a) Cd 3d and (b) S 2p for lr-CdS,sr-CdS and tp-CdS.

Fig.4 (a) Diffuse reflectance spectra of lr-CdS,sr-CdS and tp-CdS and (b) corresponding Kubelka-Munk model showing the band gap values of 2.54,2.51 and 2.50 eV,respectively.

Fig.5 Photoluminescence spectra of lr-CdS,sr-CdS and tp-CdS.

We then tested the BET surface area of three CdS samples(Table 1).The results show the specific surface area of tp-CdS,sr-CdS and lr-CdS gradually increased from 19.8 to 35.9 and to 78.1 m2·g-1.The ratio of surface to bulk tends to increase.Since surface atoms are usually coordinatively unsaturated,they are prone to the formation of defect sites or surface reconstruction.This is consistent with the INBE/IDEsequence in the PL spectroscopy results.

Table1 Comparison of BET surface area and photocatalytic hydrogen production rate (HPR) over different CdS nanocrystals.

3.2 Photocatalytic hydrogen production test

We next performed the photocatalytic hydrogen production reaction over above three CdS crystals.Table 1 compares the hydrogen production rate over different catalysts.lr-CdS exhibits the highest photocatalytic activity,with a hydrogen rate of 482 μmol h-1·g-1,which is 2.6 times that over sr-CdS(183 μmol·h-1·g-1) and 8.8 times that over tp-CdS (55 μmol·h-1·g-1).To better study the effect of exposed crystal plane on the activities,we try to normalize hydrogen production rate on a basis of unit surface area to rule out the influence of the surface area.The normalized hydrogen production rates are 6.17,5.10 and 2.78 μmol·h-1·m-2for lr-CdS,sr-CdS and tp-CdS,respectively.It is reported that rodlike CdS crystals mainly expose nonpolar surface (100),while the plate-like CdS can expose more 002 surface (polar).In our case,the proportion of nonpolar faces increases with the increase in the aspect ratio from tp-CdS to sr-CdS and to lr-CdS.So,the nonpolar surface shows greater contribution to the activity than does the polar surface.However previous work has suggested that the polar surface has higher reactivity than nonpolar one,especially for ZnO in methanol synthesis reaction,photocatalysis,etc.It seems that our results contradict the traditional concept.In combination with the aforementioned PL spectra,we find that the variation of hydrogen production rate show a similar trend as the amount of surface defects (lr-CdS ＞ sr-CdS ＞ tp-CdS).It is speculated that the activity can be associated with the surface defects which may serve as active sites for hydrogen evolution.One may argue that a better crystallization usually results in a higher photocatalytic activity since the defects in the bulk crystal would serve as the sites for the photo-induced charge recombination.A good crystallization often means there are very few lattice defects in the crystal,but this is not much associated with surface defects.The crystallization degree from XRD or TEM cannot actually reflect the degree of surface defects.As we know,surface defects,not perfect surfaces,are required for the chemical adsorption of reactants on a catalyst.It is universally acknowledged that the catalytic activity sites are more often existed in the surface or nearsurface defects because of its high surface unsaturation and high reactivity.

Based on above analysis,we propose a mechanism for the photocatalytic hydrogen production,as described in Fig.6.Under visible light irradiation,semiconducting CdS can absorb photons and electrons at the valence band (VB) are excited to the conduction band (CB),leaving holes behind at VB.The pairs of photo-induced electrons and holes are formed.The hot electrons in the bulk can migrate to the crystal surface and are trapped by the surface defects.Protons at the catalyst surface are then reduced by the reactive electrons to form H2.Accordingly,the holes can also migrate to the surface defects as well and capture electrons from the sacrificial agent.The surface defects facilitate the effective separation of photo-induced electron-hole pairs,which may account for the improvement in photocatalytic activity33–36.

Fig.6 Proposed mechanism on surface defects induced enhancement of hydrogen production rate over CdS catalysts.

3.3 Effect of Pd deposition

Generally,CdS itself is a good photon-capture material with visible light response,but it still suffers from low hydrogen production rate24,25,30.This is because the fast recombination of photo-induced electron-hole pairs and the low redox reaction rate by consumption of photo-induced charges.Although the surface defects can assist the separation of electrons and holes,but it is still far from enough,because the defect sites show high potential for hydrogen production.The fabrication of the heterojunction structure is normally used to solve the low activity of semiconducting catalysts such as TiO2and CdS.Cocatalysts,not only favor the effective separation of photo-induced charges,but also accelerate the surface redox reaction via decreasing activation energies37–43.It is generally believed that Pt-group noble metals are the best catalysts for hydrogenation reaction and hydrogen production reaction25.So in this work,we next introduced Pd to CdS surfaces by using surface defects as anchoring sites for the deposition of Pd.

The deposition of Pd was realized by in situ deposition of 1%PdS onto CdS,just prior to the photocatalytic reaction.After addition of Pd,all catalysts show significantly increased hydrogen production rate,as seen in Table 1.Specifically,sr-CdS/Pd complex shows the best performance,with a H2rate of 7884 μmol·h-1·g-1,which is comparable to ever-reported advanced values in the literature.We also obtained increased activities over lr-CdS/Pd (2873 μmol·h-1·g-1) and tp-CdS/Pd(589 μmol·h-1·g-1).When compared to unpromoted CdS,the reaction rate is greatly enhanced by 43.1,10.7 and 6.0 times over sr-CdS,lr-CdS and tp-CdS,respectively.Therefore,the tiny amount of Pd has great promotion to CdS in hydrogen production.Such high activity may result from easy separation of electron/hole pairs and decreased activation energy required for hydrogen evolution.

The sr-CdS/Pd catalyst,as an example,was then characterized using TEM and XPS to study the interfacial structure and chemical state of Pd.Fig.7a shows a high resolution TEM image of sr-CdS/Pd.It shows that the edge of CdS nanorod is decorated with PdS NPs with particle size of 2–3 nm.Note that surface defects on CdS may play an important role for the construction of Pd-CdS heterojunction,thus facilitating the photo-induced electron/hole transfer at the interfaces.Fig.7b shows the Pd 3d XPS spectrum.The main peak at 336.6 eV is assigned to PdS species,while a small peak at 335.0 eV is attributed to metallic Pd(0).A peak at 338 eV can be assigned to PdOx species,probably due to the re-oxidization of Pd after exposure in air.XPS quantitive analysis shows a Pd/Cd molar ratio of 0.02,consistent with recipe Pd dosage (1%,w).This confirms the successful deposition of Pd on CdS crystal.

Fig.7 (a) TEM image and (b) high resolution XPS spectrum of Pd 3d region for the PdS/sr-CdS nanocomposite.

Fig.8 Proposed mechanism on the Pd-CdS heterojunction induced enhancement of hydrogen production rate.

At last,we believe that great promotion of Pd can be explained by traditional synergetic catalysis mechanism,as shown in Fig.8.It is reported that part of PdS surface can be reduced to metallic Pd state by photo-induced electrons at PdS-CdS interfaces30.The formed Pd0would decrease the potential for hydrogen evolution reaction,thus accelerating the hydrogen production.Meanwhile,the PdS can accept the holes from the VB of CdS,and further promote the charge separation degree.Therefore,the heterojunction is crucial to the high photocatalytic hydrogen production rate.

4 Conclusions

In summary,we have compared three CdS nanocrystals with different morphologies as photocatalysts for hydrogen production reaction.With the increase in the aspect ratio from tp-CdS to sr-CdS and lr-CdS,the area of exposed nonpolar surface increases,and the degree of surface defects increases as well.It is found that lr-CdS shows a higher hydrogen production rate than does sr-CdS and tp-CdS,and the activity is related to the degree of surface defects.Surface defects can effectively trap photo-induced electrons/holes,thus decreasing their probability of recombination.The defect sites may serve as active sites for hydrogen production.In addition,these defects can be used to anchor Pd particles to form heterojunction structure,which assists the separation of photo-induced charges.With the assistance of 1% Pd,the hydrogen production rate has been greatly enhanced over all CdS catalysts.Notably,the hydrogen production rate over sr-CdS/Pd reaches 7884 μmol·h-1·g-1,which is comparable to ever-reported advanced values in the literature.Hopefully,this work provides knowledges on the understanding of crystal surface and heterojunction structure interfaces for the improvement in the utilization of solar energy.