Performance of ZnO dye-sensitized solar cells with various nanostructures as anodes

合集下载

Epitaxial Growth of ZnO Nanodisks with Large Exposed Polar Facets on Nanowire

1© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, WeinheimEpitaxial Growth of ZnO Nanodisks with Large Exposed Polar Facets on Nanowire Arrays for Promoting Photoelectrochemical Water SplittingH aining C hen ,Z hanhua W ei ,K eyou Y an ,Y ang B ai ,Z onglong Z hu ,T eng Z hang ,a nd S hihe Y ang *1.I ntroducti on P hotoelectrochemical water splitting is an important way to producing hydrogen, which has been attracting intense attention because of the pressing need to perpetuate a sustainable society with energy security and a greenenvironment.[1] To generate hydrogen efﬁ ciently and stably DOI: 10.1002/smll.201401298S ingle-crystalline and branched 1D arrays, ZnO nanowires/nanodisks (NWs/NDs) arrays, are f abricated to signiﬁ cantly enhance the per f ormance o f photoelectrochemical (PEC) water splitting. The epitaxial growth of the ZnO NDs with large exposed polar facets on ZnO NWs exhibits a laminated structure, which dramatically increases the light scattering capacity of the NWs arrays, especially in the wavelength region around 400 nm. The ND branching of the 1D arrays in the epitaxial fashion not only increase surface area and light utilization, but also support fast charge transport, leading to the considerable increase of photocurrent. Moreover, the tiny size NDs can facilitate charge separation and reduce charge recombination, while the large exposed polar facets of NDs reduce the external potential bias needed f or water splitting. These advantages land the ZnO NWs/NDs arrays a f our times higher power conversion ef ﬁ ciency than the ZnO NWs arrays. By sensitizing the ZnO NWs/NDs with CdS and CdSe quantum dots, the PEC perf ormance can be further improved. This work advocates a trunk/leaf in forest concept for the single-crystalline NWs/NDs in array with enlarged exposure of polar f acets, which opens the way for optimizing light harvesting and charge separation and transport, and thus the PEC water splitting.NanowiresD r. H. Chen, Z. Wei, Dr. K. Yan, Y. Bai, Z. Zhu, T. Zhang, Prof. S. YangD epartment of Chemistry William Mong Institute of Nano Science and TechnologyT he Hong Kong University of Science and Technology Clear Water Bay K owl oon ,H ong Kong E-mail :c hsyang@ust.hkin photoelectrochemical cells (PEC), metal oxide semi-conductors, such as TiO 2 [2] and ZnO, [3]have been exten-sively explored as photoelectrodes. Nanocrystals of these materials are particularly relevant because their high sur-face-to-volume ratio would shorten the diffusion length of photocarriers. However, there is a large amount of grain boundaries in nanocrystalline photoelectrodes, which has been commonly considered as important carrier recombina-tion sites.[1g,2e,4] Besides, these boundaries would also lower the electron transport rate and worsen the recombinationlosses.[1g ,2i ,3f ,4a ]I n order to reduce electron recombination and enhance water-splitting performance, considerable efforts have been invested to explore new photoelectrode architectures. One dimensional (1D) nanostructure arrays, such as nanorods or nanotubes directly grown on substrate, are thought to be an optimal architecture because of their fewer crystal boundaries and effective charge transport along the directionsmall 2014,DOI: 10.1002/smll.201401298H. Chen et al.© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheimfull papersperpendicular to the substrate.[ 1g ,2d ,2i ,3c ,3d ,3f ,5] However, most of these 1D arrays-based photoelectrodes show poor power conversion performance because of their relatively low sur-face area for carrier transfer (due to their generally large sizes) and low visible light harvesting ability (due to theirpoor light scattering capacity).[ 2d ,2i ,3f ]B ranching 1D arrays could be an effective approach to resolving the above problem. For example, Zheng et al. branched TiO 2 nanorods with tiny nanorods to improve the electron-hole pair separation, charge transfer and visible light utilization, which hence enhanced the water-splittingperformance.[ 2e ] In our previous work, ZnO nanotetrapods were also branched with small nanorods to improve the PECperformance.[ 3g ] However, most of the previously reported branched 1D arrays have grain boundaries between trunks and branches. In other words, the trunks and branches are incoherent crystalline domains falling short of the mono-crystallinity, which to some extent also need to be considered for further accelerating electron separation/transport and reducing carrier recombination losses.I n this work, single-crystalline, branched nanowire arrays were successfully constructed through the epitaxial growth of ZnO nanodisks (NDs) on ZnO nanowire (NWs) arrays. This nanostructure is not only desirable for increasing surface area and enhancing light scattering but also suitable for fast electron transport. Also, the tiny size of ZnO NDs favors the separation of electron-hole pairs so as to suppress the recom-bination loss. Indeed, we have demonstrated that with the large exposed polar facets of NDs, the external potential bias for water splitting was largely reduced. Consequently, the ZnO NWs/NDs photoelectrode reaped a four times higher PEC water splitting efﬁ ciency than the non-branched ZnO NWs counterpart. In addition, the PEC performance was considerably improved after co-sensitization with CdS and CdSe quantum dots (QDs).2.R esults and Di scussi on 2.1.G rowth of ZnO NWs/NDs Arrays Atwo-step hydrothermal (HT) method was applied to grow single-crystal ZnO NWs/NDs (see the Experimental Section for more details), which is illustrated in F igure 1A . First, ZnO seeds were coated on an FTO glass by spin coating a sol. After sintering, the ZnO seeds-coated FTO glass was immersed inthe ﬁ rst HT (1 st HT) growth solution containing ZnCl 2,hexa-methylenetetramine (HMT), NH 3·H 2O and polyethylenimine. The XRD pattern in Figure S1a (Supporting Information) indicates that all diffraction peaks are indexed to be hex-agonal ZnO structure with the preferential growth direction along [0001] (or [002]). The ZnO NWs grown at this step is about 13–14 µm in length and 100–200 nm in diameter (Figure S2A, Supporting Information). Finally, ZnO NDs were grown on ZnO NWs by simply immersing NWs-coated FTOglass in the second HT (2nd HT) growth solution containing Zn(NO 3)2, HMT and sodium citrate. [ 6] And the reaction was conducted at 95 °C for 24 h. After the secondary growth, noany new diffraction peak is presented and the peak intensity of (002) planes increases more obviously than those of others, suggesting that the [0001] direction of ZnO NDs is also ver-tical to substrate (see Figure S1b, Supporting Information).F igure 1 B portrays the overall cross-sectional mor-phology of the ﬁ nal NWs/NDs array. Overwhelmingly, it can be likened to a forest and the individual NWs/NDs has sim-ilar relation of trunk to leaf for a tree. Compared with bare ZnO NWs (Figure S2A, Supporting Information), the sec-ondary growth of ZnO NDs coarsens the NW surfaces. All of the NWs are well covered with NDs which have a narrow size distribution and seem to uniformly grow on the NWs with a laminated structure. The close-up cross-sectional SEM image in Figure 1 C clearly shows the orderly protruding NDs on the NWs with a vivid laminated structure, and strikingly, all of the NDs are vertical to the NWs. On a closer examina-tion, the ND thickness is about tens of nanometers and the inter-ND distances are also very small, resulting in a large increase of the surface area compared with the non-deco-rated NWs. The roughness factors of the NWs/NDs and NWs, deﬁ ned as the ratio of the total surface area of ZnO to its projected FTO substrate area, were estimated to be 363 and 219, respectively. Growth of NDs on NWs should result from the distribution of surface defects on the NWs, which couldinduce the secondary growth of ZnO.[ 7] To support this prop-osition, we found that the density of NDs on NWs decreased (see Figure S3, Supporting Information) after the NWs were annealed at elevated temperatures due to the diminution ofthe surface defects.[ 8] V iewed from top (Figure 1 D ), the NWs/NDs display a beautiful ﬂ ower-like structure. The petals, corresponding to the NDs, surround the NWs from all directions, which increase the diameter of the NWs from 100–200 nm to 400–600 nm, and thus the lateral diameter of the NDs is cal-culated to be about 150–250 nm. The presence of ﬂ ower buds in the centre may be due to the slight further growth of the ZnO NWs along the [0001] direction.T EM was also used to probe the ﬁ ner details of the ZnO NWs/NDs. First, from the low-magniﬁ cation TEM images in F igure 2 A ,B, we can see that the NDs with a thickness of about 20–40 nm are pinned up vertically to a high precision on the NW stalks. Second, such NWs/NDs are single-crystal-line structure in whole from the spotted electron diffraction pattern. This claim is well supported by the coincident lattice fringes (such as, typical (0001) planes) of NDs in the high-resolution TEM (HRTEM ) images (Figure 2 B 1,2), taken from the areas 1 and 2 in Figure 2 B , respectively), sugges-tive of the epitaxial growth of NDs from NWs. As such, the NDs have large exposed (0001) and (000–1) facets. Gener-ally, according to the intrinsic anisotropy of ZnO, the growth rate ( v ) along different directions have the following order: v [0001] >> v [01–10] > v [000–1], which commonly results in theformation of nanorod or nanowire structure.[ 9] However, in our case, citrate anions tend to absorb on the (0001) planesand form complexes with the Zn2+ ions on (0001) facets, [ 6] which greatly suppress the growth rate along the [0001] direction, that is, v [0001] < v [01–10], leading to the formationof the nanodisk or nanosheet structure.small 2014, DOI: 10.1002/smll.201401298Epitaxial Growth of ZnO Nanodisks with Large Exposed Polar Facets on Nanowire Arrays3 © 2014 Wiley-VCH Verlag GmbH & Co. KGaA, WeinheimB y carefully observing the high magniﬁ cation cross-sec-tional SEM image in Figure 1C and TEM image in Figure 2 A , we ﬁ nd that all the ends of ZnO NDs are not ﬂ at but show a wedge-shape structure. Figure 1 C , 2 A combined indicate that the tips tapered bottom up, resulting in the smooth top (000l) facets but the slightly coarser bottom (000–1) facets. This is a signpost that the growth rate along the [000–1] direction is higher than that along the [0001] direction, in keeping with the preferred absorption of citrate ions on the (0001) planes to the (000–1) planes. Therefore, in the secondary growth system, the growth rate along the different directions have the order of v [01–10] > v [000–1] > v [0001].Interestingly, an individual ND broken off from a NW during the TEM sample preparation was caught in the image, as presented in Figure 2 C . An incomplete hexagonal structure can still be distinguished from this individual ND as it was at its initial growth stage. The HRTEM image in Figure 2 C 1 shows that the lattice fringes can be indexed to the (100) planes, conﬁ rming the lateral growth of ZnO NDs along the direction vertical to the [0001] axis. Plausibly, the hexagon-shape structure of the NDs is a consequence of their epitaxial growth from the hexagon-shape structure of the NWs. Summarizing the above, Figure 2 D illustrates the initial growth feature of ND on NW.small 2014,DOI: 10.1002/smll.201401298F igure 1. A ) Schematic illustrating the preparation processes of ZnO NWs/NDs, B) Large scale, C) close-up cross-sectional, and D) top-view SEM images of the as-prepared ZnO NWs/NDs.H. Chen et al.© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheimfull papers2.2.I nﬂ uence of the Sodium Citrate Concentration on ZnO NWs/NDsDuring the growth of ZnO NDs, sodium citrate was used as a capping agent to block the growth of ZnO along with [0001] direction in favor of the ZnO ND product. Capping effects, which are directly related to the concentration of sodium cit-rate, affect the morphology of ZnO NDs and optical proper-ties of the ZnO NWs/NDs arrays, and therefore are studied in some detail.F or bare ZnO NWs without secondary growth, the sur-face of NWs seems to be smooth and the diameter is about 100–200 nm, as shown in F igure 3 A . The top-view in the inset of Figure 3 A shows a typical hexagonal structure of ZnO crystal. After secondary growth in the solution containing 50 mg/L of sodium citrate, the surface of NWs becomes coarse (see Figure 3 B ) and their diameter obviously increases. The change of ND thickness with the concentration of sodium citrate is calculated and presented in Figure 3 F . At the concen-tration of 50 mg/L, the ND thickness is about 660 ± 240 nm. From the top-view morphology (inset in Figure 3 B ), the hex-agonal structure is still maintained, but their size obviouslyincreases to about 200–300 nm.W ith the increase of sodium citrate concentration to 100 mg/L, the ND thickness obviously decreases to about 130 ± 45 nm, see Figure 3 (C,F), and the hexagonal structure can no longer be observed but becomes rounded (inset in Figure 3 C ). This is because that higher lateral growth ratewill lead to the elimination of corresponding planes.[ 10 ]In addition to the change in shape, the NDs present different growth orientations around the centre of NWs, resulting in the formation of a ﬂ ower-shape structure. The ﬂ ower buds can also be observed on the top of NWs. When the concen-tration increases up to 250 mg/L, the capping block effect on (0001) facets was further enhanced with the ND thickness decreasing to about 38 ± 8 nm, as shown in Figure 3 D ,F, assmall 2014,DOI: 10.1002/smll.201401298F igure 3. I nﬂ uence of the sodium citrate concentration on the morphology of ZnO NWs/NDs: A) bare NWs, B) 50 mg/L, C) 100 mg/L, D) 250 mg/L,and E) 500 mg/L. Insets are the top-view morphology of the corresponding samples with a scale bar of 300 nm. F) The relationship of ND thicknesswith the sodium citrate concentration.F igure 2. T EM characterization of ZnO NWs/NDs: A–C) TEM images; B1,B2,C1) HRTEM images corresponding to area 1 in (B), area 2 in (B), and area 1 in (C), respectively. Inset in (B) is the corresponding electron diffraction pattern. D) Schematic illustrating the growth feature of ZnO ND on ZnO NW.Epitaxial Growth of ZnO Nanodisks with Large Exposed Polar Facets on Nanowire Arrays5 © 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheimwell as in Figure 1 , 2 presented above. Correspondingly, the diameter of ﬂ ower-shape structure increases, and both the sizes of ﬂ ower petals and buds increase.F rom the analysis above, furthering the capping effect by increasing the concentration of sodium citrate to 500 mg/L is expected to decrease the ND thickness even more. Figure 3 E ,F bears out this prediction. Sure enough, the ND thick-ness was decreased to about 17 ± 5 nm. However, the diame-ters of the ﬂ owers, the sizes of the ﬂ ower petals and buds also decreased, contrary to what we have expected. We realized then, that in addition to the capping effect through the for-mation of complexes with the Zn2+ ions on the (0001) planes to block the longitudinal growth of NW, citrate anions couldalso form complexes with Zn2+ ions in the growth solution, [ 11 ] and thus reduce the amount of Zn2+ ions to sustain growth of ZnO, limiting the sizes of NDs.S ince the NW array only grows along the direction ver-tical to the substrate, light passing through it could not be effectively utilized because of the inter-NW gaps. When NDs are grown perpendicularly on the NWs, light irradiated on the gaps are better utilized by the laminated structure, which either absorb the light or scatters the light to increase the effective optical path length. In order to evaluate the optical property of NWs and NWs/NDs, diffuse reﬂ ectance spectra were recorded. Compared with NWs, as shown in F igure 4A , all the NWs/NDs demonstrate considerably a higher diffuse reﬂ ectance, especially in the short wavelength region. Evi-dently, the optical property of the NWs/NDs is closely tied to their structure. As the diameter of the ﬂ ower-shape struc-ture increases by increasing the sodium citrate concentration from 50 mg/L to 250 mg/L, the diffuse reﬂ ectance is gradually enhanced over almost the whole wavelength range. But the diffuse reﬂ ectance decreases when the concentration further increases up to 500 mg/L. Expressing the diffuse reﬂ ectance against the sodium citrate concentration, Figure 4 B clearly exposes this change trend in the range of 450 and 600 nm, which can be incontrovertibly attributed to the inter-NW gaps being ﬁ lled with laminated NDs and thus the higherscattering capacity of the photoelectrode.T he high light scattering capacity at the wavelengths near or below 400 nm is very important for ZnO photoelectrodes to efﬁ ciently harvest the light because the characteristic absorption of ZnO is around 400 nm. This suggests that theZnO NWs/NDs arrays grown at 250 mg/L would most efﬁ -ciently utilize the irradiated light most efﬁ ciently for PEC. Therefore, the application of the ZnO NWs/NDs (250) arrays in PEC is further explored in detail below.2.3.P EC Performance of ZnO NWs/NDs Z nO NWs/NDs (250) was ﬁ rst investigated for PEC using the ZnO NWs sample as a reference. A standard three-electrode system was used with ZnO NWs/NDs or NWs, Pt wire and Ag/AgCl electrode as the working, counter and reference elec-trodes, respectively. The electrolyte was an aqueous solutioncontaining 0.5 m Na 2S O 4(pH = 6.5). As shown inF igure 5 A , obvious photocurrent was detected under light illumination for both photoelectrodes. And as expected, NWs/NDs givesmuch higher photoelectric conversion than NWs.T hrough a careful comparison between the J –V curves of the two photoelectrodes, three different features could be obviously distinguished. First, NWs/NDs has a signiﬁ cantly higher photocurrent density in the whole potential range than NWs. The larger surface area (about 66% enlarge-ment) and obviously increased light utilization for NWs/NDs could be an important explanation. Compared with those bare ZnO-based photoelectrodes (without sensitiza-tion or doping) reported in literatures,[3a ,3b ,3d ,3f ,3g ] the pho-tocurrent density obtained by the present ZnO NWs/NDs is promising.S econd, upon sweeping the potential anodically, the J –V curve of NWs/NDs exhibits a steeper increase in current with respect to potential and achieves a saturated photocurrent density at 0.7 V versus Ag/AgCl, whereas no an obvious satu-ration of the photocurrent density is detected for NWs. The achievement of the saturated photocurrent density is striking, which was rarely observed for ZnO-based photoelectrodeseven when catalysts were used. [ 3a ,3b ,3d ,3f ,3g ] This feature sug-gests that the much more efﬁ cient charge separation and col-lection in NWs/NDs.[ 2e ] Compared with NWs, NDs with much smaller size guarantees the effective transport of the holes to the ZnO/electrolyte interface, thereby reducing electron-holerecombination.[ 2e ] A large fraction of the polar facets of ZnO NDs might also favor the separation of electron–hole pairsand reduce recombination.[ 12 ] small 2014,DOI: 10.1002/smll.201401298F igure 4. A ) Diffuse reﬂ ectance spectra of the ZnO NWs/NDs arrays with the NDs grown in solutions containing different sodium citrate concentrations.B) Relationship of diffuse reﬂ ectance values at 450 nm and 600 nm with the sodium citrate concentration.H. Chen et al.© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheimfull papersThird, NWs shows the photocurrent onset potential at –0.28 V versus Ag/AgCl, a typical values for the previouslyreported ZnO photoelectrodes,[ 3a ,3b ,3d ,3f ,3g ] while it negatively shifts about 0.11 V to –0.39 V versus Ag/AgCl for NWs/NDs, as is further shown in Figure 5 B . The negative shift of the photocurrent onset potential in J –V curve means lower external potential bias is needed for water splitting. There-fore, without a doubt, the much higher photoelectrochemical water splitting activity of NWs/NDs in comparison with the reference NWs has been borne out.T he power conversion efﬁ ciency (PCE,η ) for a water splitting photoelectrode that requires an applied voltage canbe assessed using the following equation:[ 1a ] (1.23)/app light I V P η=− (1) w here V app is the applied voltage versus reversible hydrogen electrode (RHE), I is the externally measured current den-sity, and P light is the power density of the illuminating light. The potential was measured against Ag/AgCl reference elec-trode and converted to RHE potential by using the Nernstequation: [ 3g ]0.059pH 0.1976V RHE Ag/AgCl =++E E (2)T he result is displayed in Figure 5 C . The maximum PCEvalue for NWs/NDs is 0.10% (at 0.84 V vs RHE), about 4 times higher than that for NWs (0.025% at 0.95 V vs RHE). No enough photocurrent was generated at low input biases to be efﬁ cient, while higher biases negate the advantageof efﬁ cient utilization of light since the limiting voltagefor water splitting is 1.23 V vs SHE.[ 2e ] The fact that NWs/NDs exhibits a peak maximum at a lower bias demonstrates another advantage of NWs/NDs over NWs.C ompared with the photocurrent density obtained under white light illumination, incident-photon-to-current-conver-sion efﬁ ciency (IPCE) is a better parameter to characterize the photoconversion efﬁ ciency of different photoelectrodes because it gives the speciﬁ c conversion efﬁ ciencies at eachwavelength. IPCE is deﬁ ned as:[ 1g ]IPCE (1240)/()PH light I P λ=×× (3) w here I PH is the generated photocurrent density (A/m 2), λis the incident light wavelength (nm), P light is the photon ﬂux (W/m 2 ), and 1240 is the unit correction factor. Here, IPCE spectra of NWs/NDs and NWs were measured at 0.0 V versus Ag/AgCl. As can be clearly observed in Figure 5 D , the IPCE value for NWs/NDs is obviously higher than that for NWs in the wavelength region from 365 to 460 nm. Speciﬁ cally, the IPCE value for NWs/NDs is about 5.35% at 380 nm (cor-responding to the intrinsic band gap of ZnO), whereas the IPCE value for NWs is only 0.38% at the same wavelength. The IPCE results tally with the photocurrent difference of the J –V curves above. That is, NWs/NDs exhibits much more superior power conversion efﬁ ciency than NWs. In addi-tion, compared with the IPCE spectrum of NWs, an obvious red shift is observed for NWs/NDs. As shown in the inset of Figure 5 D , the onset of IPCE spectrum for NWs is about 420 nm, while the onset extends to about 440 nm for NWs/small 2014,DOI: 10.1002/smll.201401298F igure 5. A ) J –V curves of ZnO NWs/NDs and ZnO NWs obtained under dark (D) and light illumination (L) (1 sun, 100 mW/cm2)conditions. B) Zoomed J –V curves at low potentials. C) Relationship of PCE values with the potential vs RHE. D) IPCE spectra (inset is the zoomed IPCE spectra in the wavelength region from 380 to 450 nm).Epitaxial Growth of ZnO Nanodisks with Large Exposed Polar Facets on Nanowire Arrays7 © 2014 Wiley-VCH Verlag GmbH & Co. KGaA, WeinheimNDs. Since both NWs and NDs are only composed of ZnO, the red shift behavior can be attributed to the higher light scattering capacity for NWs/NDs (see Figure 4 ).T o gain a deeper insight into the PEC performance dif-ference between the two photoelectrodes, more compara-tive studies on the electrical properties were conducted. Theelectron diffusion coefﬁ cients ( D n) for ZnO NWs/NDs and ZnO NWs were determined using the intensity modulated photocurrent spectroscopy (IM PS) technique. The wave-length of the irradiation light for IMPS measurement is 370 nm that could surely excite and generate electrons in ZnO. At the same current density (see F igure 6A ), the NWs/NDs has a lower but very close D n in comparison with NWs. It is well known that electron transport in metal oxide nano-structrues is trap-state-limited.[ 13 ] Thus, the higher density of surface states associated with the larger surface area for branched 1D arrays than non-branched 1D array would cer-tainly lower their electron transport rate.[ 14 ] Here, the close D n for these two nanostructures demonstrates that single-crystalline ZnO NWs/NDs arrays large maintained the fast electron transport property of ZnO NWs. Furthermore, when keeping the same light irradiation intensity (see Figure 6 B ),the D n for NWs/NDs becomes signiﬁcantly higher than that for NWs, implying that the much higher light utilization by NWs/NDs could generate many more electrons to occupy thetrap states and hence further increase D n. Therefore, both the single-crystalline ZnO NWs/NDs arrays and its higher lightutilization can improve the charge transport rate.F urthermore, ﬂ at band potentials ( V FB ) at the ZnO/electrolyte interface were measured through M ott-Schottky ( M –S ) plots to explain the negative shift of the onset poten-tial in J –V curve for ZnO NWs/NDs. The results are shown inFigure 6 C , and the relevant parameters can be extracted fromﬁ tting to the M –S equation: [ 15 ]1/(2/)[()/]200d FB 0C e N V V kT e εε=−− (4)w here e 0 is the electron charge,ε the dielectric constant of ZnO, ε 0 the permittivity of vacuum, N d the dopant density,V the electrode applied potential, V FB the ﬂ at band poten-tial, and k T /e 0 is a temperature-dependent correction term. Extrapolation of the M –S plots gives intercepts at the X axis and thus the respective V FB values. To our expectation, the V FB for ZnO NWs was found to be about –0.44 V , and it is negatively shifted to about –0.68 V for ZnO NWs/NDs, which naturally explains the negative shift of the photocur-rent onset potential for NWs/NDs. Plausibly, the negative shift of V FB can be attributed to the prevalence of the polar facets for NWs/NDs, which tend to effectively adsorb anionssuch as OH −ions [16] and thereby uplift the conduction band edge.[17] 2.4.Q Ds-Sensi ti zed ZnO NWs/NDs for PEC To further improve the PEC based on ZnO NWs/NDs, quantum dots (QDs) were deposited to increase light utili-zation. The sensitizing processes of CdS and CdSe QDs are illustrated in F igure 7 A . Firstly, NWs/NDs was successivelyimmersed in Cd2+ ion solution and S 2− ion solution for many times to be sensitized with CdS QDs, which was commonly called successive ion layer absorption and reaction (SILAR)method.[ 18 ] Then, CdSe QDs were also deposited on CdS QDs-sensitized NWs/NDs by SILAR method. That is, CdSQDs-sensitized NWs/NDs was succes-sively immersed in Cd2+ ion solution and Se2− ion solution for many times. [ 19 ] T he TEM image in Figure 7 (B) shows that both the surfaces of NWs and NDs become obviously coarse, clearly conﬁ rming the deposition of QDs. The HRTEM image in Figure 7 C shows that many small particles with the sizes of about several nanometers are coated on the surface of NDs. The lattice fringes enclosed within the yellow circles corre-spond to the (111) planes of CdS (PDF 65–2887). The lattice fringes consistent with (111) or (220) planes of CdSe (PDF 19–0191) can be also easily observed, as enclosed by the red circles and red dash circles, respectively. No obvious lattice fringes corresponding to those of ZnO were observed, demonstrating that CdS and CdSe QDs can be well deposited and covered on ZnO NWs/NDs by the SILARmethod. Elemental mapping in Figure S4(Supporting Information) further testiﬁes the presence of O, Zn, Cd, S, and Se ele-ments, which seamlessly cover the surfaces of ZnO NWs/NDs.small 2014,DOI: 10.1002/smll.201401298F igure 6. E lectrical properties of ZnO NWs/NDs and ZnO NWs. A) Electron diffusion coefﬁ cient as a function of current density. B) Electron diffusion coefﬁ cient as a function of light intensity. C) M –S plots (inset is the zoomed M –S curve for NWs/NDs). All of the measurements were conducted in a three-electrode system with an Ag/AgCl reference electrode and a Pt wire counter electrode.。

Performance of LiNi1_3Mn1_3Co1_3O2_graphite batteries based on aqueous binder

Performance of LiNi1/3Mn1/3Co1/3O2/graphite batteries based on aqueous binderNicholas Loefﬂer,Jan von Zamory,Nina Laszczynski,Italo Doberdo,Guk-Tae Kim*, Stefano Passerini*Institute of Physical Chemistry&MEET Battery Research Centre,University of Muenster,Germanyh i g h l i g h t sManufacturing of CMC-based electrodes for Li-ion batteries.Successful aqueous processing of water-sensible NMC cathode material.Thermal and electrochemical stability of CMC binder.Effect of calendering on CMC-based electrodes.Superior long-term cycling performance of CMC-based Li-ion cells.a r t i c l e i n f oArticle history:Received6August2013 Received in revised form10September2013Accepted1October2013 Available online15October2013Keywords:Aqueous electrode processing Sodium carboxymethylcellulose BinderLithium-ion batteriesLiNi1/3Mn1/3Co1/3O2(NMC) Graphite a b s t r a c tThis manuscript reports on the manufacturing and characterization of sodium carboxymethylcellulose-based,Li-ion positive electrodes with high active material mass loadings using only water as a solvent. The effect of different calendering forces on the aqueous processed cathode electrodes is also reported. Finally,the performance of balanced full Li-ion cells in pouch cell conﬁguration is investigated.These Li-ion cells subjected to long-term cycling experiment displayed an average coulombic efﬁciency of99.96% and retained a speciﬁc capacity of almost70%of its initial capacity after2000cycles.Ó2013Elsevier B.V.All rights reserved.1.IntroductionFrom the electrochemical point of view Li-ion batteries consist of“active”and“inactive”components[1].At the electrode level, only the cathode and anode active materials are listed among the “active”components whereas the binder,conductive agent and current collector,which do not directly contribute to the elec-trochemical processes(i.e.,to the speciﬁc capacity of the battery), are summarized as“inactive”components.Despite the attribute “inactive”,these components may have a quite signiﬁcant impact on cell performance and can also contribute to reversible and irreversible capacities within the cell[2e4].In fact,the capacity of an electrode is inﬂuenced by the active material mass loading (mg cmÀ2)and density(mg cmÀ3).These parameters are deter-mined by the thickness of the coating,density,porosity and the active material content in the slurry[5].Especially the nature of the binder signiﬁcantly inﬂuences these parameters.This latter component is supposed to hold the electrode’s active material particles in touch with each other and the conductive additive (cohesion)and ensure a superior bonding of the composite electrode to the current collector(adhesion)[6e8].Moreover,the binder needs to buffer the volume expansions which occur during (de)lithiation of the active material particles[8e10].For being deployed in a Li-ion battery,the binder should be electro-chemically stable in a wide voltage range extending from about 0V to5V vs.Li/Liþand chemically resistant to other cell com-ponents(electrolyte).Fulﬁlling most of these requirements,*Corresponding authors.E-mail addresses:kimguktae@uni-muenster.de(G.-T.Kim),stefano.passerini@ uni-muenster.de(S.Passerini).Contents lists available at ScienceDirectJournal of Power Sourcesjournal h omepage:www.elsevier.co m/lo cate/jp owsour0378-7753/$e see front matterÓ2013Elsevier B.V.All rights reserved./10.1016/j.jpowsour.2013.10.018Journal of Power Sources248(2014)915e922polyvinylidene-di-ﬂuoride(PVdF)is nowadays the state of the art binder in commercial Li-ion batteries,especially for cathodes [11].Nevertheless it exhibits a few serious drawbacks.Like all ﬂuorinated polymers,its poor recyclability is problematic with respect to the environment.Furthermore,its use is accompanied by the utilization of the volatile,expensive and highly toxic sol-vent N-methyl-pyrrolidone(NMP).Additionally with a price of around15e18EUR kgÀ1,PVdF is a relatively costly polymer [12,13].Hence a growing interest is focussing on the substitution of PVdF with environmentally benign binders,which are water soluble.This would result in signiﬁcant process cost savings and vast reduction of toxic material use,resulting in an environ-mentally more sustainable way of producing Li-ion batteries [14,15].There is a growing spectrum of possible water soluble binders for electrode preparation comprising different kinds of binder materials, e.g.rubbers[16,17],gums[18,19],and other poly-saccharides[20,21]of which the sodium salt of carboxymethyl-cellulose(CMC)seems to be the most promising candidate [12,13,22,23].CMC is cheap(1e2EUR kgÀ1),water soluble,easy disposable and has already proven to be an adequate replacement for PVdF concerning the anode(graphite)electrode[14,24,25].Recent works showed that CMC is also feasible as a binder for different cathode materials[15,23,26,27].Nevertheless,only few results have been,so far,reported on its use in lithium-ion battery cathodes.To close this gap and further shorten the distance be-tween lab-scale experiments and industrial production,this manuscript is focused on demonstrating a way of manufacturing a CMC-based,lithium-ion cells with high cathode mass loadings,i.e., close to those used in commercial lithium-ion batteries.As lithium nickel e manganese e cobalt dioxide(LiNi1/3Mn1/3Co1/3O2or NMC)is among the presently used active cathode material[11]for com-mercial batteries,we decided to combine the NMC cathode with a graphite anode(SLP30).2.Experimental2.1.Electrode processingCommercially available NMC(T ODA,average particle size: d90¼10m m)and SLP30graphite(T IMCAL,average particle size: d90¼32m m)were used as delivered.Sodium carboxymethyl cel-lulose(CMC,D OW W OLFF C ELLULOSICS,Walocel CRT2000PPA12)with a degree of substitution of1.2was used as binder,while the con-ducting agent carbon black was C-NERGY Super C45(T IMCAL,pri-mary average particle size:30nm).The anode and cathode electrodes were prepared in the following way:CMC wasﬁrstly dissolved in deionized water at room temperature by magnetic stirring.After complete solvation of CMC the solution was trans-ferred to a steel vessel and the required amount of Super C45was added.Further stirring was conducted with a dispersing system (Dispermat VL,VMA G MB H)at medium stirring speed(2000rpm) for approximately4h.The required amount of active material was then added into the aqueous mixture and the slurry was further homogenized at20 C for3h using the Dispermat dissolver.In a ﬁnal step the slurry was mixed at lower stirring speed(200rpm) under vacuum for10e30min.The so-obtained anode slurry was transferred to a pre-pilot automated coating line(H OHSEN C ORP,Coating Machine HSCM-20802i)and casted on copper foil(thickness:10m m).The coated electrode was immediately dried in the heating zone(130 C)of the coating machine in ambient air.The obtained cathode slurry was cast on aluminium foil (thickness:20m m),which was previously coated with a carbon dispersion(T IMCAL,C-NERGY Li-Quid101)to prevent corrosion of the aluminium current collector,by using a laboratory scale blade coater[28].The coated electrode was immediately pre-dried in an atmospheric oven(B INDER,ED-115)at80 C for30min.The dried anode and cathode electrodes were cut into stripes of 6cm width and roll-pressed with a calendering machine(H OHSEN C ORP,Roll Press Machine HSRP-2025)at different pressures.Viscosity measurements of all slurries were carried out at shear rates of100sÀ1and500sÀ1using a rheometer(T HERMO S CIENTIFIC, HAAKE Viscotester550).The composition of the dried cathode electrode tape was 88wt.%NMC,7wt.%Super C45and5wt.%CMC while that of the anode electrode tape was90wt.%SLP30,5wt.%Super C45and 5wt.%CMC.The average mass loadings of cathode and anode electrodes were about7.5mg cmÀ2and3.3mg cmÀ2,respectively.2.2.Cell assemblingFor the half-cell tests cathode and anode disc electrodes with an area of1.13cm2were cut out of the electrode tapes.The electrode discs were dried at170 C under vacuum for12h.The cathode half-cells were prepared in pouch bag conﬁgura-tion with metallic lithium(R OCKWOOD L ITHIUM,thickness:50m m, battery grade)as counter electrode.Aluminium tabs were used to connect the NMC electrode,while nickel tabs were used for the lithium electrode.A commercially available single layer poly-ethylene membrane(A SAHI K ASEI,Hipore SV718)was used as sepa-rator.All pouch bag half-cells were assembled in a dry-room (R.H.<0.1%)at20 CÆ1 C.The anode half-cells were prepared in the three-electrode Swagelok cell system with metallic lithium(R OCKWOOD L ITHIUM,bat-tery grade)as counter and reference electrode.A glassﬁbre sepa-rator(W HATMAN,GF/D)drenched in electrolyte was used for the Swagelok cells,which were assembled in an argonﬁlled glove box (MB RAUN,Labmaster DP)with oxygen and water contents lower than1ppm.Lithium-ion full cells were also assembled in the pouch bag conﬁguration.The cathode and anode electrodes with an active area of,respectively,16cm2and19.36cm2were cut out of the electrode tapes and dried at170 C under vacuum for12h.Nickel (anode)and aluminium(cathode)tabs were used as cell termi-nals.Asahi separator was used in the full cells,which were assembled in the dry-room by piling up anode,separator and cathode.For all the assembled cells the commercial electrolyte consisting of1mol LÀ1LiPF6EC:DMC1:1w/w(M ERCK AG,LP30) was used.2.3.Electrode characterizationAll electrochemical tests were performed with a M ACCOR Battery tester4300at room temperature(20 CÆ2 C).The cathode half-cell tests were carried out between3.0V and4.3V vs.Li/Liþwith an initial charge/discharge rate of0.1C for3cycles followed by a constant cycling at1C rate.The cycling tests of all cathode elec-trodes were performed in galvanostatic conditions(CC).The anode half-cell tests were carried out between0.02V and1.50V vs.Li/Liþwith an initial charge/discharge rate of0.1C for6cycles.Further cycling was performed at1C rate for20cycles,then increased to2C rate for additional20cycles andﬁnally followed by a constant cyclization at1C rate.All anode tests were performed as galva-nostatic and potentiostatic measurements(CCCV).The full-cell measurements were also carried out with CCCV tests in a voltage range between2.75V and4.20V.Scanning electron microscopy(SEM)imaging of the different pressed electrodes was performed using a Z EISS EVO MA10 microscope.N.Loefﬂer et al./Journal of Power Sources248(2014)915e922 9162.4.Binding agent characterization2.4.1.Thermal gravimetric analysis(TGA)Thermogravimetric measurements of the pure CMC powder were performed with a Q5000IR TGA instrument(TA I NSTRUMENTS). Platinum pansﬁlled with15e25mg of sample were heated from room temperature to1000 C with a heating rate of10 C minÀ1, using nitrogen(25mL minÀ1)or oxygen(25mL minÀ1)as purge gas.2.4.2.Cyclic Voltammetry(CV)Cyclic Voltammetry(CV)investigations of the pure CMC binder were carried out using a VMP3(B IO L OGIC).Therefore the CMC was dissolved in deionized water,casted on aluminium or copper foils with a lab scale blade coater and immediately pre-dried in an at-mospheric oven at80 C for30min.Electrodes with an area of 1.13cm2were cut out of the CMC-coated metal foils and dried at 140 C under vacuum for12h.The CV measurements were per-formed in Swagelok cells with the same set-up as the previously described anode half-cells(see2.2).The tests were carried out at a scan rate of0.5mV sÀ1and a scan range for the CMC-coated aluminium and copper foil of,respectively,1.5V e5V and0.02V e2.50V.3.Results and discussion3.1.Binder stabilityThe performance of a Li-ion battery is strongly affected by the electrochemical and thermal stabilities of the CMC binder used to make the composite electrodes.To verify the binder thermal sta-bility TGA measurements in oxygen and nitrogen were performed on CMC powder.The thermogravimetric measurements of CMC binder in oxygen as well as nitrogen atmosphere are shown in Fig.1a and b.As displayed in Fig.1a CMC binder is thermally stable in oxygen atmosphere up to200 C.At temperatures above200 Cthe material shows a sharp decrease in weight due to the thermal decomposition of the carboxy groups and,ﬁnally,the glucopyr-anose units that make up the cellulose backbone.Above260 C approximately20wt%residue is left most likely due to the for-mation of sodium containing ashes.Fig.1b illustrates the thermal stability of pure CMC powder in the nitrogen atmosphere.It is evident that CMC binder is also stable up to200 C in inert gas. Above230 C the CMC starts to decompose in several stages until at approximately800 C once more the stable sodium residues are obtained[29,30].Summarizing these data,it can be concluded that the CMC binder is not affected by thermal treatments below200 C in air,thus conﬁrming that the electrodes can be pre-dried in air at 130 C and,ﬁnally,under vacuum at170 C(see section2.2for the electrode drying procedure).Apart from the thermal stability during the electrode process-ing,the electrochemical stability of CMC binder during charge/ discharge process is also of major importance.Fig.2a and c show the results of the characterization performed on bare and CMC coated(approx.coating thickness:1m m)copper(Fig.2a and b)and aluminium foils(Fig.2c and d).To investigate the electrochemical cathodic stability of the CMC binder,cyclic voltammetry measurements on bare and CMC coated copper foils(some selected cycles are reported in Fig.2a and b, respectively)were performed in the potential range extending from0.2V to3.0V vs.Li/Liþ.The results of the CV measurements reveal no evidence of any peak related to increased currentﬂows due to the CMC binder decomposition.Actually,the CMC coating is seen to reduce the cathodic currentﬂowing in theﬁrst sweep. Hence the CMC binder is electrochemically stable in-between the potentials occurring at the anode.The peaks appearing during the ﬁrst cycle of the CV measurement may be attributed to oxygen at 1.7V and water contamination at1.1V.At lower potentials(0.02V e 1.00V)reduction peaks of solvent components as well as salt an-ions appeared.The major peak at about0.6V,however,is associ-ated to the insertion of lithium in the native copper oxide layer [31,32].Fig.2c displays the CV measurement of aluminium foil.Upon cycling the current response show the expected decrease at high potential(above4V vs.Li/Liþ).This stems from the growth of an insulating passivation layer consisting mostly of aluminiumﬂuo-ride(AlF3)on the aluminium foil due to electrolyte decomposition, which is protecting aluminium from anodic corrosion[33,34].This effect is less pronounced for the CMC-coated aluminium foil (Fig.2d).Considering the narrow current range(between7m A andÀ7m A)shown in Fig.2c and d as well as the smooth curve shapes,the CMC binder proofs also to be stable at the potentials occurring at the cathode electrode.3.2.Anode electrodesControlling the slurry viscosity is an important factor which signiﬁcantly inﬂuences the processing and,even,the electro-chemical properties of the electrodes[19,24,35].Especially when increasing the total amount of processed slurry,changing the mixing procedure from lab-scale magnetic stirring to a pre-pilot dissolver(Dispermat),and utilizing a pre-pilot automated coating line for the coating process,monitoring the viscosityis Fig. 1.TGA weight loss proﬁle of CMC powder in oxygen(a)and nitrogen(b) atmospheres.N.Loefﬂer et al./Journal of Power Sources248(2014)915e922917indispensable to ensure consistent electrode properties.The anode slurry composition showed a dynamic viscosity of 2.82Pa s at a shear rate of 100s À1and 1.03Pa s at a shear rate of 500s À1before the coating process.Since the measured viscosity of the electrode dispersion originates primarily from the CMC solution,the viscosity measurement exhibits an expected shear-thinning rheological behaviour.Preinvestigations (data not shown here)revealed that also anode slurries with a two times lower viscosity showed a good processability.Hence,there seems to be a broad viscosity range for processing the anode slurry on the pre-pilot automated coating line.To investigate the cycling performance of anode half-cells upon pressure,the manufactured anode electrodes were roll-pressed at different line force loads (kg cm À1)[36].However,since graphite materials are sensitive to high calendering forces,the electrodes were subjected only to moderate compression (up to 7kg cm À1).Table 1reports the electrode density of the compressed electrodes as well as their thickness reduction (in %of the original thickness).To verify the effect of electrode calendering,which might endanger the electrode breaking the soft graphite particles and thereby close the electrode pores,SEM characterization was per-formed.In Fig.3are compared the SEM images of un-pressed (Fig.3a),mildly pressed (3kg cm À1,Fig.3b),and heavily pressed (>5kg cm À1,Fig.3c)electrodes.As shown by the SEM images no signi ﬁcant differences are visible between the un-pressed and the mildly pressed electrodes.Both electrodes exhibit a homogenous distribution of active material particles and the pores are evenlyspread throughout the electrode surface.However,the heavily pressed electrode (Fig.3c)shows a very ﬂat surface with most of the porosity occluded.Fig.4a displays the cycling performance of differently pressed anodes (in half-cell con ﬁguration)at current densities corre-sponding to 1C and 2C rate using the CCCV testing mode.The test cells were pre-activated with 6cycles at 0.1C rate and for 9cycles at a 1C rate to favour the formation of a stable SEI at the interface.During the ﬁrst 10cycles at 1C rate the delivered capacity of all half-cells continuously increased upon cycling.This can be attrib-uted to a still not optimized electrolyte wetting of the electrode at initial cycling and is maximized in the heavily pressed electrode [12].However,it becomes apparent that the cycling of the un-pressed electrode performs as well as that of the mildly pressed (3.0kg cm À1)electrode.The heavily pressed electrodes (5.0kg cm À1and 7.0kg cm À1),however,showed a somewhat less stable cycling performance with increasing current density to 2C and a slightly higher capacity fading after the 50th cycle.Never-theless,after the 15th cycle the speci ﬁc capacity varied between 371mAh g À1and 376mAh g À1for all of the half-cells.Considering the theoretical speci ﬁc capacity of SLP 30to be 372mAh g À1,these results are quite satisfying.The capacity exceeding the theoretical value obviously originates from the added carbonaceous conduc-tive agent Super C45[37,38].Fig.4b illustrates the voltage pro ﬁle of the 3kg cm À1pressed electrode.At the 5th cycle at a current density corresponding to 0.1C rate,the well-known stages for lithium-ion intercalation in graphite are apparent [39,40].When increasing the C-rate the intercalation stages are continuously shifted towards lower po-tentials vs.Li/Li þ.Remarkably the curves of the 25th cycle (before 2C rate)and the 65th cycle (after 2C rate)are practically over-lapping.The analysis of the depicted voltage pro ﬁles supports the results of the morphological investigations and the cycling exper-iments.On the one hand,the electrochemical behaviour is not negatively in ﬂuenced by mildly pressing the anode electrodes.On the other hand,highly reversible cycling and an excellent capacity retention,even when increasing to 2C rate for twenty cycles,is achieved as con ﬁrmed by the reproducibility of the voltage pro ﬁles.Fig.2.Selected cyclic voltammograms (1st,2nd,3rd,5th,10th cycle)of cells with (a)pure Cu foil,(b)CMC coated Cu foil,(c)pure Al foil and (d)CMC coated Al foil working electrodes.Counter and reference electrodes:Li.Table 1Density and thickness reduction of anode electrodes calendered at different pressures.Calendering force (kg cm À1)Electrodedensity (g cm À3)Thickness reduction (%)00.77030.84450.931170.9918N.Loef ﬂer et al./Journal of Power Sources 248(2014)915e 922918Overall,considering that the 3kg cm À1pressed electrode shows very stable cycling behaviour,comparable with that of un-pressed electrodes,as well as a more homogenous growing of the solid electrolyte interface on a ﬂattened electrode surface (and therefore a bene ﬁcial effect on the cycling behaviour in a full cell).Hence,the mildly calendered electrode was selected to prepare the lithium-ion batteries [41].3.3.Cathode electrodeIn contrast to the described anode electrode,the aqueous pro-cessing of the cathode material is facing an additional challenge:as soon as the NMC material gets into contact with water the pH of the suspension rises up to approximately 11[42].Since aluminium,which is standardly used as current collector for the cathode electrode,is chemically not stable at this high pH value,coating of the aqueous NMC slurry results in a serious aluminium corrosion [28].The substrate degradation is obviously enhanced by the application of thick coatings (here the wet coating thickness was 220m m)to achieve high electrode mass loading,which results in a large volume of the corrosive ﬂuid on the aluminium foil.The SEM images in Fig.5a and b show the appearance of an aluminium foil previously coated with the aqueous NMC slurry,which was dried in an atmospheric oven at 80 C (the NMC coating was removed prior to SEM characterization).Consistently with the results from Ishii et al.[42]and Doberdo et al.[28],the entire surface of the aluminium foil is covered with corrosion products (Fig.5a).Some distinct corrosion pits (Fig.5b)perforating the aluminium foil were also observed.To avoid the corrosion issue and,as a consequence,the capacity losses due to formation of insulating corrosion products and damages to the aluminium current collector [28],two ways of corrosion prevention seem to be viable:First,the addition of a mild acid to the slurry prior to the coating step which results in a lower pH value and therefore prevent the aluminium foil from being corroded.Kim et al.reported that formic acid was used tohinderFig.3.SEM images of (a)as coated and (b and c)calendered at 3kg cm À1and >5kg cm À1SLP 30anodeelectrodes.Fig.4.Cycling performance (a)of SLP 30anode electrodes pressed with different line force loads and selected voltage pro ﬁles (b)of the calendered (line force load:3kg cm À1)anode electrode.Counter and reference electrodes:Li.N.Loef ﬂer et al./Journal of Power Sources 248(2014)915e 922919aluminium corrosion during the manufacturing of aqueous pro-cessed Li 4Ti 5O 12anode electrodes [12].Second,a thin carbon layer can be coated on top of the aluminium foil which protects the Al foil from the corrosive aqueous medium [28].Adding formic acid to slurries comprising NMC as the active material involves the danger of active material damaging due to leaching of transition metal ions.Table 2reports the metal con-centration obtained by ICP analysis of a formic acid e water solution in contact with NMC.Thus,the second approach,apply a carbon coating layer to protect the aluminium foil (see:2.1)was selected.Since the pre-coating of such a layer was technically dif ﬁcult toprepare at the used pre-pilot automated coating line,the coating process was carried out at a lab-scale blade coater.In order to achieve high mass loadings of the cathode electrode the viscosity of the cathode slurry has to be controlled very care-fully.The need for suf ﬁciently high cathode mass loadings arises from the need of properly balancing the electrode capacity in the full cells (described later).Because the speci ﬁc theoretical capacity of the anode active material (372mAh g À1)exceeds the speci ﬁc theoretical capacity of the cathode active material (161mAh g À1)by a factor of nearly two,a cathode with a high active material mass loading is needed for a proper balancing.To achieve high mass loadings of the cathode electrodes,an electrode slurry composition with a dynamic viscosity of 2.31Pa s at a shear rate of 100s À1and 0.98Pa s at a shear rate of 500s À1was prepared.Naturally the cathode electrode slurry exhibited the same shear-thinning rheo-logical behaviour as the anode electrode one.Adjusting the vis-cosity to lower values resulted in insuf ﬁcient mass loadings of the cathode electrode.Adjusting the viscosity to higher values led to an inhomogeneous mixing of the electrode components.Apparently,changes in viscosity seem to have a signi ﬁcantly larger in ﬂuence on the processability of the cathode slurry compared to the anodeone.Fig.5.SEM images of the (a)corroded Al foil surface and (b)of an individual pitting corrosion hole.Table 2Concentration of extracted transition metal ions from 2.64g NMC in 7mL distilled H 2O at different amounts of formic acid.Concentration of extracted ions (mg mL À1)Amount of formic acid (mg)045100180380Cobalt 00.43 1.20 1.90 4.28Nickel00.43 1.18 1.85 4.10Manganese0.481.532.032.13Fig.6.SEM images of the (a)as coated and calendered NMC cathode electrodes at line force loads of (b)17kg cm À1,(c)47kg cm À1and (d)70kg cm À1.N.Loef ﬂer et al./Journal of Power Sources 248(2014)915e 922920Hence we operate in a very narrow viscosity range when increasing the active material mass loading of the cathode electrodes.In order to investigate the in ﬂuence of different calendering forces,the cathode electrode tapes were compressed at 17kg cm À1,47kg cm À1and 70kg cm À1.In Fig.6the SEM images of these electrodes after roll-pressing and drying are depicted.In general all the electrodes show a homogenous distribution of active material particles.The SEM image of the un-pressed electrode (Fig.6a)shows very clearly the formation of extensive cracks throughout the electrode surface due to the large shrinkage of the CMC binder during the drying procedure.This is clearly due to the high average mass loading (about 7.5mg cm À2)and the limited capability of our lab-scale equipment to properly mix and coat slurries with higher viscosity.Interestingly,however,the ﬁssures diminish in size with increasing calendering force as can be seen in Fig.6b e d.A closer examination of the most compressed electrode (Fig.6d)evidence that all cracks have been closed in consequence of the applied force.Moreover the electrode surface pressed by 70kg cm À1is well ﬂattened in comparison to the un-pressed electrode (Fig.6a).Summarizing,strong calendering has a considerably bene ﬁcial ef-fect on the morphology of high mass loading CMC-based NMC cathodes.Table 3reports the electrode density of the compressed electrodes as well as their thickness reduction (in %of the original thickness).With the purpose of investigating the effect of calendering on the electrochemical behaviour of NMC cathodes cycling tests were performed on electrodes pressed at different line force loads.The results are illustrated in Fig.7a showing the cycling performance at 1C rate for these electrodes.For all the tested samples the speci ﬁc discharge capacity drops,as expected,to values between 130mAh g À1and 135mAh g À1after the initial 3formation cycles at 0.1C rate,during which capacities higher than 150mAh g À1were detected.Upon cycling at 1C rate the un-pressed electrodes show a signi ﬁcantly higher capacity fading over the illustrated 70cycles than the pressed electrodes.The differences in speci ﬁc discharge capacity of the pressed electrodes are usually ranging around 1e 2mAh g À1.It can,therefore,be considered that the compression step is needed to improve the long-term cyclability,however,independently from the line force load applied during the compression step all electrodes show the same behaviour upon cycling.Nevertheless,the electrode compressed at 70kg cm À1certainly showed the ﬂatter surface and,for such a reason,is preferable.According to the presented charge/discharge results,Fig.7b il-lustrates a few,selected voltage pro ﬁles of the 70kg cm À1pressed electrode.The sloping Li-insertion plateau typically observed for NMC [43],is observed between the cut-off potentials of 3.0V and 4.3V.Focussing on the curves recorded at a discharge rate of 1C,a very limited decrease of delivered capacity is noticed,corre-sponding to a total capacity fading of approximately 5.2%between the 4th and the 70th cycle.Considering their promising cycling performance and superior morphological appearance,the electrodes pressed at 70kg cm À1were selected for the realization of the lithium-ion cells.Table 3Density and thickness reduction of cathode electrodes calendered at different pressures.Calendering force (kg cm À1)Electrode density (g cm À3)Thickness reduction (%)0 1.49017 1.571340 1.6618701.8828Fig.7.Cycling performance (a)of cathode electrodes calendered at different pressures and selected voltage pro ﬁles (b)of the calendered cathode electrode (line force load:70kg cm À1).Counter and reference electrodes:Li.Fig.8.Speci ﬁc capacity and ef ﬁciency (a)and selected voltage pro ﬁles (b)of a full Li-ion cell comprising a high mass loading NMC cathode electrode (7.64mg cm À2,calendered at 70kg cm À1)and a SLP 30anode electrode (3.14mg cm À2,calendered at 3kg cm À1).N.Loef ﬂer et al./Journal of Power Sources 248(2014)915e 922921。

高性能氢气传感器

Polymer-assisted Deposition of Co-doped Zinc Oxide Thin Film for the Detection ofAromatic Organic CompoundsWei Li, Dojin Kim*Abstract-Cobalt-doped Zinc oxide thin film was deposited onto SiO2/Si substrate using polymer-assisted deposition method. The surface morpholog y, phase structure and chemical state of the thin film were characterized by SEM, XRD, and XPS. The g as-sensing characteristics of the thin film upon exposure to aromatic org anic compound vapors were investig ated with a home-made sensor measurement system. The current results show that the film morphology is influenced by the doping of Co, and the sensor behavior was quite different between undoped and Co-doped ZnO thin films.B ACKGROUNDZnO has been widely used as a low cost sensing element for detection and monitoring of hazardous gases and vapors. Their advantages are high sensitivity, fast response, and long-time stability. Of the family of ZnO sensor, thin film type may be more favorable structure since the gas sensing properties are determined by the gases absorbing on the materials surfaces leading to a change of charge carriers and thus, of the materials resistances [1]. In the past several years, a large number of papers about ZnO sensor have been published. However, only a few ZnO sensors could be found for detection of aromatic organic compounds (AOC) [2-4]. Considering adverse effect of AOC on environment and human, developing reliable sensors to monitor the AOCs from the surrounding atmosphere becomes more and more important.The conventional methods for preparation of zinc oxide thin films can be divided into two categories: one is the vacuum techniques and the other is wet chemical solution methods. Recently, an alternative method named polymer-assisted deposition (PAD) was developed, and was successfully implemented to deposition of metal oxide thin film [5]. This method does not suffer from either the high cost for device fabrication, or complex operation and rigorous conditioning in the processing in comparison to conventional methods.In the current work, ZnO thin films are prepared by the PAD method. Furthermore, Co-doped ZnO thin films are also prepared. We compared the gas sensing properties of the ZnOC URRENT R ESULTSThe deposition of Co-doped ZnO thin films were carried out as the following. A certain amount of cobalt sultanate, zinc acetate and polyethylenimine (PEI, branched, Aldrich) were dissolved in 30 ml water with assistance of sonication. The solution was then spin-coated (2000 rpm for 30 s) onto SiO2/Si substrates followed by annealing in a quartz tube furnace of air environment. Gas sensor device structure was completed by sputter deposition of platinum electrodes onto the films through a metal shadow mask of interdigitated pattern. For the gas sensing property measurements, a bubbler is used to deliver the AOC vapors because they are in liquid form at room temperature (Fig. 1).The SEM images of the thin films shown in Fig.2 suggest slight change in the surface morphology due to the doping of Co. The XRD patterns of the thin film shown in Fig.3 showed that only pure wurtzite structures of ZnO are detected both in the undoped and doped ZnO thin films. The chemical state of the thin film is identified by XPS (Fig.4), and the results show that Co is doped as intended. The response to m-xylene is shown in Fig. 5. Shown is that the resistance change of doped ZnO is opposite to undoped ZnO because of different carrier types. The recovery was poor in p-type ZnO:Co thin film. The related mechanism and other sensor properties will be further presented in detail.R EFERENCES[1] P.P.Sahay, R.K. Nath, “Al-doped ZnO thin film as methanol sensors”, Sens.Actuators B 134 (2008) pp.654-659.[2] B.L.Zhu, C.S.Xie, J.Wu, D.W. Zeng, A.H.Wang,X.Z.Zhao, “ Influence ofSb, In and Bi dopants on the response of ZnO thick films to VOCs”, Mater.Chem. Phys. 96 (2006) pp. 459-465.[3] B.L.Zhu, C.S.Xie, D.W. Zeng, W.L.Song, A.H.Wang, “Investigation of gassensitivity of Sb-doped ZnO nanoparticles”, Mater. Chem. Phys. 89 (2005) pp. 148-153.[4] S. Morandi, F. Prinetto, M. Di Martino, G. Ghiotti,O. Lorret, D. Tichit, C.Malagù, B. Vendemiati, M.C. Carotta, “Synthesis and characterization of gas sensor materials obtained from Pt/Zn/Al layered double hydroxides”, Sens. Actuators B 118 (2006) pp.215-220.Lin, G. E. Collis, H.Wang,A.D. Q. Li, S. R. Foltyn, “Polymer-assisted deposition of metal-oxidefilms”, Nature Mater. 3 (2004) pp. 529-532.978-1-4244-3544-9/10/$25.00 ©2010 IEEEFig. 1. Diagram of sensor measurement systemFig. 2. SEM images of typical Co-doped ZnO thin film. (a) undoped, (b)0.08 at. % Co, (c) 8.53 at. % Co, and (d) 18.91 at. % Co.Fig. 3. XRD pattern of Co-doped ZnO thin films.Fig. 4. XPS spectra of Co-doped ZnO thin film in (a) a wideenergy range and (b) the Co region.Fig.5. The sensor response to 1000 ppm m -xylene at 200 ć.( a) undoped ZnO and (b) 8.53 at.% Co doped ZnO.。

Dye-Sensitized Solar Cells

Electrochemical Impedance Spectroscopic Analysis of Dye-Sensitized Solar CellsQing Wang,Jacques-E.Moser,and Michael Gra1tzel*Laboratory for Photonics and Interfaces,Institute of Chemical Sciences and Engineering,Ecole PolytechniqueFe´de´rale de Lausanne,1015Lausanne,SwitzerlandRecei V ed:May25,2005Electrochemical impedance spectroscopy(EIS)has been performed to investigate electronic and ionic processesin dye-sensitized solar cells(DSC).A theoretical model has been elaborated,to interpret the frequency responseof the device.The high-frequency feature is attributed to the charge transfer at the counter electrode whilethe response in the intermediate-frequency region is associated with the electron transport in the mesoscopicTiO2film and the back reaction at the TiO2/electrolyte interface.The low-frequency region reflects the diffusionin the ing an appropriate equivalent circuit,the electron transport rate and electron lifetime inthe mesoscopic film have been derived,which agree with the values derived from transient photocurrent andphotovoltage measurements.The EIS measurements show that DSC performance variations under prolongedthermal aging result mainly from the decrease in the lifetime of the conduction band electron in the TiO2film.1.IntroductionDye-sensitized solar cells(DSC)present a promising alterna-tive to conventional photovoltaic devices.1-4After more thanone decade’s development,it has reached global AM1.5powerconversion efficiencies up to11%.5At the heart of the DSC isa mesoscopic semiconductor oxide film typically made of TiO2,whose surface is covered with a monolayer of sensitizer(Figure1).During the illumination of the cell,electrons are injectedfrom the photoexcited dye into the conduction band of the oxide.From there they pass through the nanoparticles to the transparentconducting oxide current collector into the external circuit.Thesensitizer is regenerated by electron transfer from a donor,typically iodide ions,which are dissolved in the electrolyte thatis present in the pores.The triiodide ions formed during thereaction diffuse to the counter electrode where they are reducedback to iodide by the conduction band electrons that have passed through the external circuit performing electrical work.Although these basic processes are well understood,a deeper comprehen-sion of the electronic and ionic processes that govern the operation of the DSC is warranted.Transient photocurrent/ photovoltage measurements,6-12intensity-modulated photocur-rent/photovoltage spectroscopy(IMPS/IMVS),6,13-22and very recently the open circuit voltage decay technique23,24have been used to scrutinize the transport properties of the injected electrons in mesoscopic film and the back reaction with redox species in electrolyte.Electrochemical impedance spectroscopy(EIS)is a steady-state method measuring the current response to the application of an ac voltage as a function of the frequency.25An important advantage of EIS over other techniques is the possibility of using tiny ac voltage amplitudes exerting a very small perturbation on the system.EIS has been widely employed to study the kinetics of electrochemical and photoelectrochemical processes including the elucidation of salient electronic and ionic processes occurring in the DSC.15,18,26-34The Nyquist diagram features typically three semicircles that in the order of increasing frequency are attributed to the Nernst diffusion within the electrolyte,the electron transfer at the oxide/electrolyte interface, and the redox reaction at the platinum counter electrode.15 However,owing to the complexity of the system,the unambigu-ous assignment of equivalent circuits and the elucidation of processes occurring on dye-sensitized mesoscopic TiO2electrode is difficult and remains a topic of current debate.The present study employs EIS as a diagnostic tool for analyzing in particular photovoltaic performance changes detected during accelerated high-temperature durability tests on dye-sensitized solar cells.A theoretical model is presented interpreting the frequency response in terms of the fundamental electronic and ionic processes occurring in the photovoltaic device.From applying appropriate equivalent circuits,the transport rate and lifetime of the electron in the mesoscopic film are derived and the values are checked by transient photocurrent and photovoltage measurements.We note that during the final stage of preparation of the present particle a paper by Bisquert et al.35appeared presenting a similar approach to electrochemical impedance investigations of the DSC and that concurs with our analysis.*To whom correspondence should be addressed.E-mail:michael.graetzel@epfl.ch.Figure1.Scheme of a dye-sensitized solar cell.14945 J.Phys.Chem.B2005,109,14945-1495310.1021/jp052768h CCC:$30.25©2005American Chemical SocietyPublished on Web07/20/20052.Theoretical Modeling of the Frequency ResponseThe DSC contains three spatially separated interfaces formed by FTO/TiO2,TiO2/electrolyte,and electrolyte/Pt-FTO.Elec-tron transfer is coupled to electronic and ionic transport.In the dark under forward bias electrons are injected in the conduction band of the nanoparticles and their motion is coupled to that of I-/I3-ions in electrolyte.Illumination gives rise to new redox processes at the TiO2/dye/electrolyte interface comprising sensitized electron injection,recombination with the parent dye, and regeneration of the sensitizer.During photovoltaic operation, this“internal current generator”drives all the electronic and ionic processes in the solar cell.31We now derive the equations describing the frequency response of the impedance at the different interfaces.2.1.I3-Finite Diffusion within Electrolyte and Electron Transfer at the Pt-FTO/Electrolyte Interface.In practical electrolytes,the concentration of triiodide is much lower than that of iodide and the latter is diffuses faster than the former ion.Hence,I-contributes little to the overall diffusion imped-ance,which is determined by the motion of I3-.The diffusion of I3-within a thin layer cell is well described by a Nernst diffusion impedance Z N.15,28Using Fick’s law and appropriate boundary conditions Z N becomeswhereωis the angular frequency and R equals0.5for a finite length Warburg impedance(FLW).Z0andτd are the Warburg parameter and characteristic diffusion time constant,respec-tively,which can be expressed bywhere R is the molar gas constant,T the temperature,F the Faraday constant,c0the bulk concentration of I3-,A the electrode area,D the diffusion coefficient of I3-,and d is the diffusion length.Because of the mesoporous character of the TiO2electrode,a modified Nernst diffusion impedance with R deviating from0.5is used to fit the transport of I3-.Nernst diffusion impedance in the Nyquist plot shows typically a straight line at higher frequency along with a semicircle at lower frequency.Fitting Z0andτd,the diffusion coefficient can be determined.The charge-transfer resistance R CT associated with the heterogeneous electron exchange involving the I3-T I-redox couple at the electrolyte/Pt-FTO interface is typically given for the equilibrium potential.From the Bulter-Volmer equation, one obtainswhere i0is the exchange current density of the reaction.The frequency response of charge-transfer impedance under small sinusoidal perturbation can be expressed as36where C d is the double layer capacitance.The charge-transfer resistance manifests itself as a semicircle in the Nyquist diagram and a peak in the Bode phase angle plot.For electrodes having a rough surface the semicircle is flattened and C d is replaced by a constant phase element(CPE).2.2.Electron Transport within the Mesoscopic TiO2Film and Electron Loss due to the Reduction of Triiodide at the TiO2/Electrolyte Interface.When a voltage modulation is applied in the dark to the mesoporous TiO2electrode of the DSC,electrons are injected and recovered during the cathodic and anodic parts of the current response.Their collection yield of recollecting the injected electrons depends on their diffusion lengthwhere D e is the diffusion coefficient andτr the lifetime of the electron within the film.The impedance due to electron diffusion and loss by the interfacial redox reaction in a thin mesoporous layer has been treated by Bisquert.31-34For the case of a mesoscopic TiO2film, the diffusion occurs over a finite length and is coupled with interfacial electron-transfer reaction.The electron charge is screened by the electrolyte,which eliminates the internal field, so no drift term appears in the transport equation.4The continuity equation contains therefore only the diffusion and reaction terms. the boundary condition beingwhere n is the concentration of electron,n0is their initial concentration,and L is the film thickness.To account for a harmonically modulated voltage,a frequency term is introduced yielding finally for the impedance response:where R d and R r are the diffusion and dark reaction impedance, respectively,whileωd′(ωd′)1/τd′)D e/L2)andωr(ωr)1/τr) are the corresponding characteristic frequencies.If R r f∞,eq 9describes a simple diffusion process within restricted bound-aries.For a DSC exhibiting a current collection efficiency close to unity,the condition R r.R d applies and eq9becomesUnder these conditions,the Nyquist plot shows a short straight line at higher frequencies due to diffusion and a large semicircle in the lower frequency regime,indicating fast electron transport and long lifetime of electron in the film.By contrast,if the electron collection efficiency is low,i.e.,a major part of the electron reacts with I3-in the electrolyte before they are recovered at the current collector,the condition R d.R r applies leading to Gerischer impedance34,37Z N )Z(iω)Rtanh(iτdω)R(1)Z)RTn2F2cA D(2)τd)d2/D(3)RCT)RTnF1i0(4)Z)iRCTi-RCTCdω(5)Ln) D eτr(6)∂n∂t)De∂2n∂x2-(n-n)τr(7)∂n∂x|x)L)0(8)Z)(R d R r1+iω/ωr)1/2coth[(ωr/ωd′)1/2(1+iω/ωr)1/2](9)Z)13Rd+Rr1+iω/ωr(Rr.Rd)(10)Z)(R d R r1+iω/ωr)1/2(R d.R r)(11)14946J.Phys.Chem.B,Vol.109,No.31,2005Wang et al.The Gerischer impedance produces response curves similar to a FLW impedance (eq 1).It shows a Warburg diffusion-like straight line at higher frequencies along with a semicircle at lower frequencies.The diffusion coefficient D e and the lifetime τr of the electron in the film can be obtained from eqs 10and 11.The above discussion was based on DSC in the dark and under forward bias.Under illumination the continuity equation becomeswhere R ′is the effective absorption coefficient,ηinj is thequantum yield for charge injection,and I is the incident photon flux,the boundary condition being again given by eq 8.As indicated in Figure 2,under illumination the short circuit photocurrent J SC of the cell iswhere J inj is the flux of injected electron and J loss is the current from back reaction loss.Since ηcc )1for distances x <L n ,and ηcc )0if x >L n ,where J inj,x <Ln is the anodic flux of electron that are injected by the sensitizer and collected at the FTO,while J FTO -TiO 2is the cathodic current flowing from the FTO into the TiO 2film.Thus,where J inj,x >Ln is the flux of injected electron,which is lost completely before arriving at FTO/TiO 2interface.It is clear that J inj,x <Ln <J FTO -TiO 2as E >V OC (Figure 2a);and J inj,x <Ln >J FTO -TiO 2as E <V OC (Figure 2c).At open circuit state (Figure 2b),J inj ,x <Ln equals to J FTO -TiO 2,and the total flux is zero.Phenomenologically we can treat the perturbation imposed by the illumination as if a voltage bias was applied in the dark.This applies in particular for a situation where the diffusion length of the electrons is commensurate with or larger than the film thickness.Hence,the same approach as above can be used to analyze its frequency response.Nevertheless,the model may overstate the diffusion rate D e because photons travel through the film faster than conduction band electrons.But the correction is small if the electrondiffusion length is long compared to the film thickness L .31Indeed,as will be shown below,in a DSC L n normally exceeds L rendering this distinction irrelevant.2.3.Equivalent Circuits and Typical Impedance Spectra of DSC.For a nanoporous electrode,the infinite transmission line is normally used as the equivalent circuit for modeling.For simplicity,only the representative elements displayed in Figure 3are employed here to model DSC at different states.From left to right,Figure 3shows the electron transport at the FTO/TiO 2interface,electron transport and electron capture by the I 3-at the TiO 2/electrolyte interface,diffusion of I 3-in the electrolyte,and charge transfer at electrolyte/Pt -FTO interface,respectively.For a cell exhibiting a carrier collection efficiency near unity,the condition R r .R d and eq 10applies.In this case,the equivalent circuit for the mesoporous TiO 2film comprises a diffusion element Z W1that is in series connected with the charge-transfer element R REC ,the two being in parallel with a capacitive (constant phase angle)element CPE3,as shown in Figure 3a.On the other hand,for cells where only a fraction of the photogenerated charge carriers are collected,the condition of R d .R r applies and a single Gerischer impedance element Z G describes the diffusion of the electron in the mesoscopic TiO 2film and their recapture by the triiodide ions in the electrolyte (Figure 3b).R FTO/TiO 2is the resistance of the FTO/TiO 2contact and CPE1is the capacitance of this interface.The latter feature,due to overlap with other processes,is not easily distinguished.Z W2is the Warburg impedance describing the diffusion of I 3-in the electrolyte,R CE is thecharge-transferFigure 2.Schematic model of photoanode at different voltages.(a)E >V OC ;(b)E )V OC ;(c)E <V OC .L is the film thickness,L n is the electron effective diffusion length,and ηcc is the charge collection efficiency of injected electron.∂n ∂t)R ′I ηinj +D e ∂2n ∂x2-(n -n 0)τr (12)J SC )J inj -J loss(13)J SC )J inj,x <L n -J FTO -TiO 2(14)J loss )J inj,x >L n +J FTO -TiO 2(15)Figure 3.Equivalent circuits of DSC.(a)a cell showing quantitative collection of photoinjected electrons;(b)a cell showing incomplete collection of electrons.Bottom line shows the interpretation of the electrical elements of the equivalent circuit.(A)electron transfer at the FTO/TiO 2interface;(B)electron transport and back reaction at the mesoscopic TiO 2/electrolyte interface;(C)diffusion of I 3-in the electrolyte;(D)charge at electrolyte/Pt -FTO interface.Dye-Sensitized Solar Cell J.Phys.Chem.B,Vol.109,No.31,200514947impedance at the counter electrode,and CPE2is the double layer capacitance at the electrolyte/Pt -FTO interface.A typical EIS spectrum for a DSC exhibits three semicircles in the Nyquist plot or three characteristic frequency peaks in a Bode phase plot.This is illustrated in Figure 4showing Nyquist plots of a N719sensitized DSC before and after thermal aging at 80°C for 2days.The response in the intermediate-frequency regime changes greatly upon aging,indicating the conversion of a Nernst to a Gerischer impedance.Apparently,the spectra can be well fitted in terms of the corresponding equivalent circuits in Figure 3.These models will therefore be employed to interpret impedance data in the following sections.3.Experimental Section3.1.Dye-Sensitized Mesoscopic TiO 2Electrode Prepara-tion and Cell Fabrication.The preparation of mesoscopic TiO 2film has been described in ref 38.The screen-printed double-layer film consists of a 10-µm transparent layer and a 4-µm scattering layer whose thickness was determined by using an Alpha-step 200surface profilometer (Tencor Instruments).A porosity of 0.63for the transparent layer was measured with a Gemini 2327nitrogen adsorption apparatus (Micromeretics Instrument Corp.).The film was heated to 500°C in air and calcinated for 20min before use.Then the still hot spots were dipped into a 2×10-4M 2-fold deprotonated cis-RuL2(SCN)2(L )2,2′-bipyridyl-4,4′-dicarboxylic acid)(N719)or cis-RuLL ′-(SCN)2(L )2,2′-bipyridyl-4,4′-dicarboxylic acid,L ′)4,4′-dinonyl-2,2′-bipyridyl)(Z907)dye (Chart 1)solution in aceto-nitrile/tert -butyl alcohol (1:1)and left for overnight.Finally,the dye-coated electrodes were rinsed with acetonitrile.For transient photocurrent/photovoltage measurements,single trans-parent TiO 2films with the thickness of 12µm were used.A sandwich cell was prepared using the dye-sensitized electrode as the working electrode and a platinum-coated conducting glass electrode as the counter electrode.The latter was prepared by chemical deposition of platinum from 0.05M hexachloroplatinic acid at 400°C.The two electrodes were placed on top of each other using a thin transparent film of Bynel polymer (DuPont)as a spacer.The empty cell was tightly held,and the edges were heated to 130°C in order to seal the two electrodes together.A thin layer of electrolyte was introduced into the interelectrode space from the counter electrode side through a predrilled hole.The hole was sealed with a microscope cover slide and Bynel to avoid leakage of the electrolyte solution.There were two electrolytes used in this paper:electrolyte 1,0.6M PMII,0.1M I 2,and 0.5M NMB in MPN;electrolyte 2,0.6M DMPII,0.05M I 2,0.5M tBuPy,0.1M LiI in AN:VN(1:1).Thermal stress tests were carried out by putting cells in an oven at 80°C and then measuring the I -V curve,impedance and transient photocurrent/photovoltage.3.2.I -V Measurements.A 450-W xenon light source (Osram XBO 450)was used as the irradiation source for the I -V measurements.The spectral output of the lamp matched the AM 1.5solar spectrum in the region of 350-750nm (mismatch <2%).Incident light intensities were adjusted with neutral wire mesh attenuators.The current -voltage character-istics were determined by applying an external potential bias to the cell and measuring the photocurrent using a Keithley model 2400digital source meter (Keithley).The overall conversion efficiency ηof the photovoltaic cell is calculated from the integral photocurrent density (J SC ),the open-circuit photovoltage (V OC ),the fill factor of the cell (ff),and the intensity of the incident light (I Ph ),3.3.Electrochemical Impedance Measurements.Impedance measurements were performed with acomputer-controlledFigure 4.Typical Nyquist plots of a N719sensitized DSC.Filled squares,fresh cell;open circles,cell after aging for 48h at 80°C.The lines show theoretical fits using the equivalent circuits shown in Figure 3a and b,respectively.The electrolyte is 0.6M PMII,0.1M I 2,and 0.5M NMB in MPN.CHART 1:Sensitizers Used in This StudyaaKey:(a)N719;(b)Z907.η)J SC ‚V OC ‚ff/I Ph(16)14948J.Phys.Chem.B,Vol.109,No.31,2005Wang et al.potentiostat(EG&G,M273)equipped with a frequency response analyzer(EG&G,M1025).The frequency range is0.005-100 kHz.The magnitude of the alternative signal is10mV.Unless otherwise mentioned,all impedance measurements were carried out under a bias illumination of100mW/cm2(global AM1.5, 1sun)from a450-W xenon light source.The obtained spectra were fitted with Z-View software(v2.1b,Scribner Associate, Inc.)in terms of appropriate equivalent circuits.3.4.Transient Photocurrent/Photovoltage Measurements. Transient photocurrent and photovoltage studies of the DSC were carried out by using weak laser pulses atλ)514nm, superimposed on a relatively intense bias illumination.The bias light was supplied by a cw450-W Xe arc lamp,equipped with a water filter and a680-nm cutoff filter.The continuous wave beam was condensed by a lens to irradiate a∼1cm2cross section of the cell,the surface of which was kept at a60°angle to the beam.The red light intensity measured at the cell position was typically120mW/cm2.The cell was oriented to expose the counter electrode side to both the bias light and laser beams. The5-ns-duration laser pulses at a wavelength of514nm were generated by a broadband optical parametric oscillator(GWU,OPO-355)pumped by the third harmonic of a30-Hz repetition rate,Q-switched Nd:YAG laser(Continuum,Powerlite7030). The laser beam was attenuated by gray filters to restrict the pulse fluence onto the cell to<100µJ/cm2.The514-nm laser light was strongly absorbed by the dye,and therefore,injected electrons were introduced into a narrow spatial region,corre-sponding to where the probe light enters the film.Current transients were measured across a20Ωresistor load using a large bandwidth digital signal analyzer(Tektronix DSA602A). Transient photovoltages were measured by feeding the signal directly into the DSA amplifier,whose impedance was1MΩ.4.Results and Discussion4.1.Impedances of DSC Obtained in Dark and Illumina-tion.There are different processes that occur in the cell in the dark or under illumination.At open circuit voltage and in sunlight,there is no net current flowing through the cell.All the injected electrons are recaptured by I3-before being extracted to the external circuit.Meanwhile,the oxidized dye is regenerated by I-.As a result,the absorbed photon energy is converted to heat through the two coupled redox cycles involving sensitized electron injection,dye regeneration,and electron recapture by I3-.The counter electrode is kept at equilibrium,because there is no net current flowing through it. However,in the dark under forward bias,electrons are transported through the mesoscopic TiO2network and react with I3-.At the same time,I-is oxidized to I3-at the counter electrode.The net current density can be large depending on the applied bias voltage.Figure5shows the impedance spectra of a DSC measured at OCV(-0.68V)under1sun and under forward bias(-0.68 V)in the dark.Strikingly,the impedance due to electron transfer from the conduction band of the mesoscopic film to triiodide ions in the electrolyte,presented by the semicircle in intermedi-ate-frequency regime,is much smaller under light than in the dark even though the potential of the film is the same. Correspondingly,the characteristic frequency shown in Bode phase plots increases two times,suggesting the electron lifetime is shortened by a factor of2.This can be ascribed to a difference in the local I3-concentration.Under illumination,I3-is formed “in situ”by dye regeneration at the mesoporous TiO2/electrolyte interface,whereas in the dark,I3-is generated at counter electrode and penetrates the mesoporous TiO2films by diffusion.As indicated by eq17,the higher local I3-concentrationproduced in the porous network under light is expected toaccelerate the recapture of conduction band electrons andshortens their lifetime within the TiO2film.Here J r is the I3-reduction current,k r is the rate constant of thereduction reaction,and c ox is the concentration of I3-;theexponentsγand are the reaction orders for I3-and electrons,respectively.4.2.Impedance of DSC with Different Electrolytes.Electrolytes exert a great influence on the photovoltaic perfor-mance of the DSC by effecting the kinetics of electronic or ionicprocesses.For instance,acetonitrile(AN)-based electrolytes havemuch lower viscosity compared with3-methoxypropionitrile(MPN),the kinetics of dye regeneration,I3-f I-reaction at counter electrode,electron transport within the TiO2film,andI-/I3-diffusion in electrolyte being faster in the former case.Consequently,much better photovoltaic performance has beenachieved.In addition,additives in the electrolyte are of greatimportance for optimization and stabilization of the TiO2/dye/electrolyte interface.TBP,39NMB,40and recently guanidiniumsalts5have been shown to be effective in increasing thephotovoltage without greatly reducing the photocurrent.Figure6shows the impedance spectra of a Z907sensitizedcell with two kinds of electrolytes at different light intensity.From the Bode phase plots,the electron lifetime with electrolyte2is much longer than that obtained with electrolyte1at thesame light intensity.According to Frank et al.,it is believedthat Li+in electrolyte2plays an important role for the longlifetime of the electron.11The characteristic time constants ofelectron transport and back reaction are obtained by fitting thespectra with the equivalent circuit shown in Figure3a.Theelectron diffusion rate D e of the cell with electrolyte2is2×10-4cm2/s at1sun,3times higher than that obtained fromIMPS by Peter et al.in an AN-based electrolytes.16,22That ofelectrolyte1is1.1×10-4cm2/s,close to the value obtainedfrom photocurrent transient measurements.12From eq18,theeffective diffusion length L n of the conduction band electronsis calculated to be∼16.2µm for MPN-based electrolyte1at1sun and at open circuit voltage.That of electrolyte2is∼30.1 Figure5.Impedance spectra of a Z907cell measured at OCV(-0.68 V),1sun or at-0.68V in dark.(a)Bode phase plots;(b)Nyquist plots.Electrolyte1is used.Jr)ekrcoxγ(n -n)(17)Dye-Sensitized Solar Cell J.Phys.Chem.B,Vol.109,No.31,200514949µm.The latter being much larger than the film thickness,all photogenerated electrons will be collected.In addition,the charge-transfer impedance at the counter electrode with electrolyte 2is much smaller than that of electrolyte 1,which is in accordance with the lower exchange current density for the iodide/triiodide couple in the latter electrolyte.A light intensity effect on the electron lifetime is apparent in both electrolytes.Under illumination,V OC can be expressed as 41with k 1and k 2being,respectively,the kinetic constant of backreaction of injected electrons with triiodide and recombination of these electrons with oxidized dye and n 0being the concentra-tion of accessible electronic states in the conduction band.Neglecting the loss term due to recombination with the oxidized dye molecules,V OC depends logarithmically on the inverse concentration of I 3-and increases with incident photon flux I .The V OC of the cell with electrolyte 2are 0.614,0.665,and 0.681V at 0.1,0.5and 1sun,respectively.Those with electrolyte 1are 0.651,0.707,and 0.725V,respectively,following the predicted logarithmical relation.Any variations of the local triiodide concentration due to light illumination appear to have a small effect.For comparison,transient photocurrent/photovoltage measure-ments were also performed.Figure 7a shows the transient photocurrent curves of Z907cell with two electrolytes at short circuit state,presenting a multicomponent process.The char-acteristic time constant τc can be fitted to one major single-exponential decay process,where the diffusion coefficient D eis estimated from eq 20.8The fitted τc values are 1.56and 1.87ms,and for a 12-µm-thick film,the calculated D e are 3.9×10-4and 3.3×10-4cm 2/s for electrolytes 1and 2,respectively.These electron diffusion coefficients are higher than those obtained from the EIS measurement.This is probably due to the fact that the bias light intensity used for these measurements was larger than 1sun,giving a higher electron concentration within the film and consequently a faster electron diffusion rate;9Also,because the fast back reaction accelerates the decay process,the value for the MPN-based electrolyte obtained from transient photocurrent measurement is overstated (J SC for electrolyte 2is 12.3mA/cm 2,whereas that of electrolyte 1is only 7.9mA/cm 2).Figure 7b shows the transient photovoltage curves of the same cells at open circuit.The fitted time constant τR for electrolyte 1is ∼3.1ms,whereas because of the large capacitive current,τR cannot be obtained for the cell with the AN-based electrolyte.The lifetime obtained from transient photovoltage measurement is again shorter than that from the EIS measurement for the same reasons that have been given above.After correcting the capacitive charging effects,electrolyte 2has a longer lifetime than electrolyte 1in agreement with the EISmeasurements.Figure 6.Impedance spectra of Z907cells at different light intensity,0.1sun (green),0.5sun (red),and 1sun (black).(a)Bode phase plots;(b)Nyquist plots.Electrolyte 1,open circles;electrolyte 2,filled squares.The inset of (b)shows the spectra at 1sun for bothelectrolytes.Figure 7.Transient photocurrent (a)and photovoltage (b)of Z907cells with electrolyte 1(blue line)and 2(black line).The red lines are the corresponding computer fits for the decay processes.The inset equations show time constants for the exponential decay.D e )L 2/(2.35τc )(20)L n )Lτrτd ′)Lωd ′ωr(18)V OC )RTF ln (AI n 0k 1[I 3-]+n 0k 2[D +])(19)14950J.Phys.Chem.B,Vol.109,No.31,2005Wang et al.。

Dye-Sensitized Solar Cells-2010 Chem. Rev.

Dye-Sensitized Solar CellsAnders Hagfeldt,*,†,‡,|Gerrit Boschloo,†Licheng Sun,‡,|Lars Kloo,‡and Henrik Pettersson⊥Department of Physical and Analytical Chemistry,Uppsala University,Box259,SE-75105Uppsala,Sweden,Department of Chemistry,KTH-Royal Institute of Technology,Teknikringen30,SE-10044Stockholm,Sweden,State Key Laboratory of Fine Chemicals,DUT-KTH Joint Education and Research Centre on Molecular Devices,Dalian University of Technology(DUT),Dalian116012,China,and Swerea IVF AB,Box104,SE-43122Mo¨lndal,SwedenReceived October30,2009Contents1.Introduction65962.Sun,Energy,and Solar Cells65983.Overview of Performance,Materials,andOperational Principles65994.Operational Principles66024.1.Overview of the Different Electron-TransferProcesses66024.1.1.Reactions1and2:Electron Injection andExcited State Decay66034.1.2.Reaction3:Regeneration of the OxidizedDyes66034.1.3.Reaction4:Electron Transport through theMesoporous Oxide Film66034.1.4.Reactions5and6:Recombination ofElectrons in the Semiconductor withOxidized Dyes or Electrolyte Species66034.1.5.Reaction7:Reduction of ElectronAcceptors in the Electrolyte at the CounterElectrode66034.2.Energetics66034.2.1.Energy Levels in Semiconductors66044.2.2.Energy Levels of Redox Systems inSolution66054.2.3.Energy Levels of Excited Molecules66064.3.Photoinduced Electron Injection vs DirectRecombination-Reactions1and266074.3.1.Electron Injection Studies on ModelSystems:Dyes Adsorbed on TiO266074.3.2.Injection and DSC Device Efﬁciency66084.4.Regeneration of the Oxidized Dyes:Reaction366094.5.Electron Transport in MesoporousSemiconductor Electrodes:Reaction466104.6.Recombination of Electrons in theSemiconductor with Oxidized Dyes orElectrolyte Species:Reactions5and666124.7.Transport of the Redox Mediator and Reactionsat the Counter Electrode:Reaction766135.Materials Development66135.1.Nanostructured Metal Oxide Electrodes66135.1.1.TiO266145.1.2.ZnO66155.1.3.Other Metal Oxides6616 5.2.Dyes66165.2.1.Metal Complexes66175.2.2.Porphyrins and Phthalocyanines6620anic Dyes66235.2.4.The Anchoring of the Dye on the OxideSurface6631bining Sensitizers66325.2.6.Summary:Development of Dyes for DSC6632 5.3.Electrolytes and Hole Conductors66335.3.1.Liquid Redox Electrolytes66335.3.2.Gel and Polymer Electrolytes66345.3.3.Ionic Liquid Electrolytes66355.3.4.Additives66365.3.5.Alternative Redox Couples66365.3.6.Solid Organic Hole Conductors66375.3.7.Inorganic Solid Hole Conductors6638 5.4.Surface Passivation of Dye-Sensitized TiO26638 5.5.Counter Electrodes66395.5.1.Platinized Conducting Glass66395.5.2.Carbon Materials66395.5.3.Conducting Polymers66395.5.4.Cobalt Sulﬁde66395.6.Tandem Cells and p-Type DSCs66396.Characterization Techniques6641 6.1.Efﬁciency Measurements,I-V,IPCE,and APCE6641 6.2.Electrochemical Methods66426.2.1.Cyclic Voltammetry,Differential PulseVoltammetry,and Square Wave Voltammetry66426.2.2.Electrochemical Impedance Spectroscopy66426.2.3.Spectroelectrochemistry6642 6.3.Photoelectrochemical Methods66426.3.1.Electron Transport Measurements66426.3.2.Electron Lifetime Measurements66436.3.3.Electron Concentration Measurements66436.3.4.Measurements of the Electron Quasi-FermiLevel66446.3.5.Charge Collection Efﬁciency and DiffusionLength66446.3.6.Photoinduced Absorption Spectroscopy66447.Module Development/DSC Modules6645 7.1.Module Designs and Performance66457.1.1.Sandwich Z-Interconnected Modules66457.1.2.Sandwich W-Interconnected Modules66457.1.3.Sandwich Current-Collecting Modules6646*To whom correspondence should be addressed.E-mail:Anders.Hagfeldt@fki.uu.se.†Uppsala University.‡Royal Institute of Technology.|Dalian University of Technology(DUT).⊥Swerea IVF AB.Chem.Rev.2010,110,6595–6663659510.1021/cr900356p 2010American Chemical SocietyPublished on Web09/10/20107.1.4.Monolithic Serial-Connection Modules66477.1.5.Monolithic Current-Collecting Modules66477.2.Accelerated and Outdoor Module Testing66477.3.Manufacturing Processes66487.4.Discussion:Future Outlook for the DifferentDSC Module Designs66498.Future Outlook66509.Acknowledgments665110.Appendix.Content Added after ASAP Publication665111.Note Added after ASAP Publication665212.References6652 1.IntroductionAt the end of last century,the possibility to use devices based on molecular components for the construction of a robust large-scale solar electricity production facility seemed utopic.But the seminal paper by O’Regan and Gra¨tzel in 19911spurred researchers to take on the challenge.With the development of dye-sensitized solar cells(DSCs),2conven-tional solid-state photovoltaic technologies are now chal-lenged by devices functioning at a molecular and nanolevel. Record efﬁciencies of up to12%for small cells and about 9%for minimodules,promising stability data,passing,for example,the critical1000h stability test at80°C with a durable efﬁciency of8-9%,and means of energy-efﬁcient production methods have been accomplished.The prospect of low-cost investments and fabrication are key features. DSCs perform also relatively better compared with other solar cell technologies under diffuse light conditions and at higher temperatures.DSCs offer the possibilities to design solar cells with a largeﬂexibility in shape,color,and transparency.Integration into different products opens up new commercial opportunities.Besides the exciting possibilities of using DSCs for solar energy application,the riddles of the device are as thrilling. How does it work?DSCs should according to the photo-voltaic textbooks in the early1990s simply not work.The paradigm was to use highly pure semiconductor materials avoiding defects and interfaces and to rely on a built-in electricalﬁeld to separate photogenerated electron-hole pairs.DSCs,in contrast,were based on a huge internal interface prepared in a simple laboratory environment without strict demands on the purity of the materials.It was a mystery how the DSC could work in the absence of a built-in electric ﬁeld.The initial research developed a relatively simple picture of how DSCs operate,reviewed in this journal in 1995.3The basic characteristics and conceptual models, reviewed in several recent articles,4-11have been reasonablysuccessful to describe various reactions and interactions.With time,however,the chemical complexity of the DSC device has become more and more evident.The DSC is a good example of a molecular system where the function of the overall device is better than predicted from the sum of the properties of its components.12There are complex interac-tions between the device components,in particular,at the oxide/dye/electrolyte interface,but the interactions also depend on external variables such as solar irradiation, temperature,and device working conditions.Also inherent in the devices are multiscaling properties,both in time and in length,which need to be characterized and handled for the optimization of the overall device performance.DSC research groups have been established around the world with biggest activities in Europe,Japan,Korea,China,and Australia.Theﬁeld is growing fast,which can be illustrated by the fact that about two or three research articles are being published every day.In Figure1,a simple and limited literature search illustrates the growth of the number of research papers and patents over the last years.The industrial interest in DSCs is strong with large multinational companies such as BASF,Bosch,and Corus in Europe and Toyota,Sharp,Panasonic,Sony,Fujikura,and Samsung in Asia.A large volume production line has been set up in the company G24i,Wales.Research companies such as Dyesol,Australia,Solaronix,Switzerland,and Peccell,Japan,are expanding,focusing on selling material components and equipment.The principle of DSCs has also become a part of the core chemistry and energy science teaching and research.Text Anders Hagfeldt is professor in Physical Chemistry and the Dean of Chemistry at Uppsala University.He obtained his Ph.D.in1993at Uppsala University and was a postdoctoral fellow with Prof.Michael Gra¨tzel (1993-1994)at EPFL,Switzerland.His research focuses on physical chemical characterization of mesoporous electrodes for different types of optoelectronic devices,speciﬁcally dye-sensitized solar cells.He has about 200scientiﬁc publications and8patent applications.He is a member of the Royal Swedish Academy of Engineering Sciences(IVA),Stockholm, and a visiting professor at the Royal Institute of Technology,Stockholm, at Dalian University of Technology,China,and at the Institute for Materials Research and Engineering in Singapore.He is the director of Center for MolecularDevices.Gerrit Boschloo obtained his Ph.D.in1996at Delft University of Technology(The Netherlands).He held postdoctoral positions at University College Dublin(Ireland)with Prof.D.J.Fitzmaurice and at Uppsala University(Sweden)with Prof.Anders Hagfeldt.After a period as a researcher at the Royal Institute of Technology in Stockholm,he currently holds a position as an Associate Professor at the Department of Physical and Analytical Chemistry,Uppsala University.His main research interest is advanced photoelectrochemical characterization of dye-sensitized solar cells and nanostructured semiconductors.He is author of more than70 peer-reviewed articles.6596Chemical Reviews,2010,Vol.110,No.11Hagfeldt et al.books have sections or chapters dealing with DSC.13,14Laboratory kits have been developed for educational purposes under the slogan “make your own solar cell”.Not only energy science but also photochemistry,photoelectrochem-istry,materials science,and transition metal coordination chemistry have signiﬁcantly beneﬁtted from DSC research.The challenge to review the DSC research ﬁeld is the speed of publications of new research data.With the risk of missing or leaving out interesting work,as well as studies ofimportance for the development of the ﬁeld,we aim to cover the DSC research in a broad sense.We will,however,limit ourselves to sensitization by molecular dyes and will not discuss sensitization by semiconductor quantum dots.This topic was recently reviewed elsewhere.15,16After some brief notes on solar energy in general and DSC in particular (sections 2and 3),we go through theoperationalLicheng Sun obtained his Ph.D in 1990in Dalian University of Technology (DUT),China.After postdoctoral stays of one year (1992-1993)with Dr.Helmut Go ¨rner in Max-Planck-Institut fu ¨r Strahlenchemie,Mu ¨lheim an der Ruhr,Germany,and two years (1993-1995as Alexander von Humboldt fellow)with Prof.Dr.Harry Kurreck in Institut fu ¨r Organische Chemie,Freie Universita ¨t Berlin,he went to Stockholm ﬁrst as an Assistant Professor (1995-1999)in Department of Chemistry,Royal Institute of Technology (KTH),and then Associate Professor (1999-2004)in Department of Organic Chemistry,Stockholm University.He moved back to KTH as a full Professor in Molecular Devices in 2004.He is now also a distinguished professor in DUT.His research interests focus on solar energy conversion at the molecular level including bioinspired systems for catalytic water oxidation and hydrogen production,artiﬁcial photosyn-thesis,dye-sensitized solar cells,supramolecular photochemistry,and light-driven water splitting.He has published more than 200peer-reviewedpapers.Lars Kloo received his Ph.D.in Inorganic Chemistry at Lund University,Sweden,in 1990and moved to his current position as Professor in Inorganic Chemistry at the Royal Institute of Technology (KTH)in Stockholm in 1998.During 1991-1995,he was in shorter or longer periods visiting scientist in other laboratories in the U.K.,U.S.A.,and New Zealand.He has been involved in DSC research for about 10years with a main focus on electrolytes,in particular ionic liquids,and more recently the electrolyte -electrode interaction.Apart from photoelectrochemical solar cells,his research interests also embrace fundamental aspects of inorganic synthesis,cluster chemistry,and computational biologicalchemistry.Henrik Pettersson is project leader at Swerea IVF AB,a research institute providing applied R&D to bring new technologies and new methods into practical applications.He has been involved in the development of dye-sensitized solar cells for more than 15years,specializing in module technologies,process methods,up-scaling,and reliability testing.Prior to joining Swerea IVF AB,he worked for Ekologisk Energi AB,one of the ﬁrst companies to work on dye-sensitized solar cells,and for four years for Professor Michael Gra ¨tzel at EPFL inSwitzerland.Figure 1.Number of publications published per year obtained from a simple and limited literature search using the keywords “dye-sensitized”and “solar”:(a)number of research articles (data source,ISI Web of Knowledge);(b)number of patent families (data source,esp@cenet).Dye-Sensitized Solar Cells Chemical Reviews,2010,Vol.110,No.116597principles of DSC(energetics and kinetics)in section4.The development of material components is treated in section5 and some speciﬁc experimental techniques to characterize DSC in the following section.Section7deals with the current status of module development,andﬁnally we end up with a brief future outlook.2.Sun,Energy,and Solar CellsIt is clear that access to economically viable renewable energy sources is essential for the development of a globally sustainable society.The mean global energy consumption rate was13TW in the year2000.Assuming a kind of “business-as-usual”scenario with rather optimistic but reasonable assumptions of population growth and energy consumption,the projection is28TW in2050for the globalenergy demand.17,18Solar energy,besides fusion,has the largest potential to satisfy the future global need for renewable energy sources.19From the1.7×105TW of solar energy that strikes the earth’s surface,a practical terrestrial global solar potential value is estimated to be about600TW. Thus,using10%efﬁcient solar farms,about60TW of power could be supplied.The sun emits light with a range of wavelengths from the ultraviolet and visible to the infrared.It peaks in the visible, resembling the spectrum of a blackbody at a temperature of 5760K.It is,however,inﬂuenced by atmospheric absorption and the position of the sun.Ultraviolet light isﬁltered out by ozone,and water and CO2absorb mainly in the infrared making dips in the solar spectrum at900,1100,1400,and 1900nm(H2O)and at1800and2600nm(CO2).When skies are clear,the maximum radiation strikes the earth’s surface when the sun is directly overhead,having the shortest path length through the atmosphere.The path length is called the air mass(AM)and can be approximated by AM)1/cos , where is the angle of elevation of the sun.The standard solar spectrum used for efﬁciency measurements of solar cells is AM1.5G(global),giving that )42°.This spectrum is normalized so that the integrated irradiance(the amount of radiant energy received from the sun per unit area per unit time)is1000W m-2.The irradiance varies depending on the position of the sun,orientation of the Earth, and sky conditions.One also distinguishes sunlight in direct or diffuse light.The direct component can be concentrated, which increases the solar cell efﬁciency by increasing cell voltage outputs.Diffuse light arises by scattering of the sunlight in the atmosphere.This fraction is around15%on average20but larger at higher latitudes and in regions with a signiﬁcant amount of cloud cover.Materials with rough surfaces such as DSCs are relatively better suited for diffuse light than perfectlyﬂat surfaces and are less sensitive to movements of the sun.The AM1.5G solar radiation spectrum can be found from different sources.21The spectrum is shown in Figure2as the irradiance of the sun as a function of wavelength.In the diagram,we also indicate the maximum current (short-circuit condition)for a solar cell device converting all incident photons below the absorption onset wavelength into electric current(see ref22).For example,the maximum short-circuit current(J sc)for a solar cell with an absorption onset of800nm is26mA cm-2.The overall solar-to-electrical energy conversion efﬁciency,η,for a solar cell is given by the photocurrent density measured at short-circuit (J sc),the open-circuit photovoltage(V oc),theﬁll factor of the cell(FF),and the intensity of the incident light(P in). Theﬁll factor can assume values between0and less than 1and is deﬁned by the ratio of the maximum power(P max) of the solar cell per unit area divided by the V oc and J sc according toThe maximum power is obtained as the product of the photocurrent and photovoltage at the voltage where the power output of the cell is maximal.Another fundamental measure-ment of the performance of a solar cell is the“external quantum efﬁciency”,which in the DSC community is normally called the incident photon to current conversion efﬁciency(IPCE).The IPCE value corresponds to the photocurrent density produced in the external circuit under monochromatic illumination of the cell divided by the photon ﬂux that strikes the cell.From such an experiment the IPCE as a function of wavelength can be calculated from where e is the elementary charge.IPCE values provide practical information about the monochromatic quantum efﬁciencies of a solar cell.Solar cell production has grown at about30%per annum over the past15years.The conventional solar cells of today, theﬁrst generation solar cells,are based on silicon.The estimated total installed capacity in2007was7.8GW.23,24 Most of this was grid-connected with a price of around$7 per peak Watt(W p).A selling price of$2/W p,which would correspond to a production cost of about$0.5/W p,would make PV competitive with electricity production from fossil fuels.Additional2.26GW was installed in2007alone(a 50%increase over the preceding year)and90%of that was installed in four countries:Germany,Spain,Japan,and the U.S.A.In2008,the worldwide photovoltaic installations increased by 6.0GW.However,the photovoltaic(PV) industry is largely dependent on governmental subsidies. Silicon-based systems make up around90%of the current PV market.The production cost is presently around$3/Wp Figure2.Photonﬂux of the AM1.5G spectrum at1000W m-2 (ASTM G173-03),and calculated accumulated photocurrent.η)JscVocFFPin(1)FF)P max/(J sc V oc)(2) IPCE)Jsc(λ)eΦ(λ))1240Jsc(λ)[A cm-2]λ[nm]P in(λ)[W cm-2](3)6598Chemical Reviews,2010,Vol.110,No.11Hagfeldt et al.but is highly dependent on the price of the silicon material. China is now the world leader in producing crystalline silicon (c-Si)based PV cells and modules with a capacity of over 2.3GW/year.The International Energy Agency(IEA)have estimated the energy payback time of c-Si PV modules, incorporated as a grid-connected roof-top installation,as between1.5and2years.25The second generation solar cells,for example,amorphous silicon,CIGS,and CdTe,are based on thinﬁlm technologies. They are becoming a competitive class of PVs,doubling production from2006to2007.The advantages of thinﬁlm solar cells include the ease of manufacture permitting a reduction of the production cost to about$1/W p,a wider range of applications with attractive appearance,and pos-sibilities of usingﬂexible substrates.The most established thin-ﬁlm technology is amorphous silicon(a-Si).26The efﬁciencies are lower than c-Si,but it has other advantages such as a lower temperature coefﬁcient for power loss.The cost is only slightly lower than that of c-Si mainly because of expensive manufacturing equipment.After an initial efﬁciency loss,the so-called Staebler-Wronski effect,stable laboratory efﬁciencies of13.5%and module efﬁciencies around5-6%have been achieved.For CdTe laboratory efﬁciencies of16.7%and module efﬁciencies of10.9%have been reported.The company First Solar has announced a production cost below$1/W p for their CdTe solar modules. The drawbacks include the toxicity and low abundance of the materials,temperature-dependent efﬁciencies,and only an average light tolerance.Nonetheless,CdTe is a highly interesting technology for future PVs.Another such technol-ogy is CIGS(copper indium gallium diselenide).CIGS systems are capable of high efﬁciencies(19.9%in the lab,27 16.7%in submodules,28and13.5%in modules.29They are very durable,and the materials cost can be kept low even though indium is rare.Since the bandgap can be tuned,they are also suitable for tandem cells.Several plants with10-50 MW capacity are now installed globally.Bothﬁrst and second generation solar cells are based on single junction devices.Calculated thermodynamic efﬁciency limits in single junction solar cells(31%)assume that absorption of an individual photon results in the formation of a single electron-hole pair and that all photon energy in excess of the energy gap is lost as heat.This so-called Shockley-Queisser limit can be overcome by the use of various types of so-called third generation solar cell devices.30 In principle,sunlight can be converted to electricity at an efﬁciency close to the Carnot limit of95%.Various schemes to achieve efﬁciencies above31%include tandem cells,hot carrier cells,multiexciton generation,multiband cells,and thermophotovoltaics.30The goal for the third generation solar cells is to deliver electricity at a large scale competitive price, that is,less than$0.5/W p.This means very effective solar cells that are produced by techniques that permit facile mass production.The impact on economics if such concepts could be implemented would be enormous,making PVs one of the cheapest options for future energy production.DSC can be considered to be a technology between the second and third generation solar cells.It has the potential to become a third generation technology utilizing the nanoscale properties of the device.In the present stage,the technology offers the following selling points:•low production cost and particularly interesting much lower investment costs compared with conventional PV technologies•design opportunities,such as,transparency and multi-color options(building integration,consumer products, etc.)•ﬂexibility•lightweight•feedstock availability to reach terawatt scale•short energy payback time(<1year)•enhanced performance under real outdoor conditions (relatively better than competitors at diffuse light and higher temperatures)•bifacial cells capture light from all angles •outperforms competitors for indoor applications3.Overview of Performance,Materials,and Operational PrinciplesThe“Bob Beamon like”-increase in efﬁciency for DSC type solar cells reported in the famous1991Nature paper by O’Regan and Gra¨tzel was bewildering.The use of a mesoporous semiconductor electrode with a high internal surface area led to a paradigm shift in theﬁelds of photoelectrochemistry and photovoltaics in general.Attempts to develop dye-sensitized photoelectrochemical cells had been made before,5,31-33but the basic problem was the belief that only smooth semiconductor surfaces could be used.The light-harvesting efﬁciency for a monomolecular layer of dye sensitizer,even for phhtalocyanines and porphyrins,which have among the highest extinction coefﬁcients known,absorb far less than1%of the AM1.5G spectrum.34Attempts to harvest more light by using multilayers of dyes were in general unsuccessful.Indications of the possibilities to increase the roughness of the semiconductor surface so that a larger number of dyes could be adsorbed directly to the surface and simultaneously be in direct contact with a redox electrolyte had also been reported before1991.For example, Matsumura et al.35and Alonso et al.36used sintered ZnO electrodes to increase the efﬁciency of sensitization by rose bengal and related dyes.But the conversion yields from solar light to electricity remained well below1%for these systems. In addition,the dyes used in these systems were unstable. Gra¨tzel,Augustynski,and co-workers presented results on dye-sensitized fractal-type TiO2electrodes with high surface area in1985.37For DSC,there was an order-of-magnitude increase when O’Regan and Gra¨tzel in1991reported efﬁciencies of7-8%.1With regards to stability,a turnover number of5×106was measured for the sensitizer(a trimeric ruthenium complex38,39).This was followed up by the introduction of the famous N3dye(structure2in Table1) giving efﬁciencies around10%.40Since the initial work in the beginning of the1990s,a wealth of DSC components and conﬁgurations have been developed.Perhaps a key concept for the future success of DSC in this regard is“diversity”.At present,several thousands of dyes have been investigated,as well as hundreds of electrolyte systems and numerous types of mesoporous ﬁlms with different morphologies and compositions.Whereas nature has over billions of years developed highly sophis-ticated molecular systems that give life on earth,chemists are at present facing the intriguing challenge of developing technologies based on complex molecular interactions that can assist in keeping the earth livable.Such technologies are expected to progress rapidly,be it through design of new materials based on fundamental insights or by statistical, combinatorial approaches.For DSC at present,in the ofﬁcial table of world record efﬁciencies for solar cells,the recordDye-Sensitized Solar Cells Chemical Reviews,2010,Vol.110,No.1165996600Chemical Reviews,2010,Vol.110,No.11Hagfeldt et al. Table1.Collection of Some Representative Ru-Complex Photosensitizersis held by the Sharp company in Japan at 10.4%(0.3%.41A criterion to qualify for these tables is that the solar cell area is at least 1cm 2.For smaller cells,a certiﬁed conversion efﬁciency of 11.1%has been reached using the black dye as sensitizer.42Early on in DSC research,the classical dyes N3and its salt analogue N719(structure 4in Table 1)and the black dye (N749,structure 3in Table 1)were developed.The certiﬁed record efﬁciencies are obtained with these dyes,and their IPCE spectra are shown together with the molecular structures of the dyes in Figure 3.22The IPCE values are close to 80%for both sensitizers across the visible part of the solar spectrum.The N3dye starts to absorb light at around 800nm,whereas the photocurrent onset is red-shifted for the black dye to 900nm.As seen from Figure 3,however,the IPCE increases only gradually from the absorption onset to shorter wavelengths due to relatively low extinction coefﬁcients of these sensitizers.There is thus much to be gained by developing sensitizers with high extinction coefﬁcients particularly in the near-IR region of the solar spectrum.The highest photocurrent densities of the black dye under AM 1.5G condition is close to 21mA cm -2,42,43which can be compared with the theoretical maximum photocurrent of 33mA cm -2from Figure 2for a dye with an absorption threshold of 900nm.Improving the solar light absorption in the 650-950nm domain (note the dip in the solar irradiance spectrum at 950nm,Figure 2)would then be one of the main directions to take aiming for DSC efﬁciencies above 15%.The other direction is to increase the photovoltage by replacing the conventionally used I -/I 3-redox couple with a system having a more positive redox potential.These are key challenges facing present DSC research and will be further discussed below.During the last 2-3years,the advent of heteroleptic ruthenium complexes furnished with an antenna function has taken the performance of the DSC to a new level.Two examples of these dyes are Z991and C101,structures 21and 28,respectively,in Table pared with the classical DSC Ru dyes,their extinction coefﬁcients are higher and the spectral response is shifted to the red.The solar cellefﬁciency of these types of sensitizers has increased continu-ously over the last 2years,and at present efﬁciencies of 12%have been reported.22There are sometimes arguments that DSCs have not developed much since the breakthrough in the early 1990s.The record efﬁciencies reached a plateau at 10-11%,and the most efﬁcient devices have remained essentially unchanged from its original concepts.It is,however,a complex system.The realization of the need to handling complexities with the plethora of material compo-nents to explore make the 12%efﬁciency now reached with one of the new classes of dyes a strong indicator for further progress in efﬁciency performance.Another essential per-formance indicator is long-term stability.Here,the progress during the last 15years has been steady,and accelerated durability tests have been passed with higher and higher efﬁciencies.Most of the earlier work has been reviewed.44-46As an example of recent results,the C101sensitizer maintains outstanding stability at efﬁciency levels over 9%under light soaking at 60°C for more than 1000h.47This is achieved by the molecular engineering of the sensitizer but also very importantly by the use of robust and nonvolatile electrolytes,such as ionic liquids,and adequate sealing materials.Good results on overall system endurance have been reported for several years,and these results are presently being conﬁrmed under real outdoor conditions.Because of the direct relevance to the manufacturing of commercial products,little work is published on these issues,and it is,for example,difﬁcult to ﬁnd information on processing issues,as well as sealing materials and methods.The industrial development and commercialization of DSCs was the topic of the DSC-IC 3conference in Nara,Japan,April 2009.Many encouraging results were presented giving conﬁdence that the DSCs can match the stability require-ments needed to sustain outdoor operation for many years.Still,many research and development activities need to be performed,and in particular,more data on outdoor ﬁeld tests is required.One important research topic is to develop protocols for accelerated long-term stability tests relevant to DSC technologies.A schematic of the interior of a DSC showing the principle of how the device operates is shown in Figure 4.The typical basic conﬁguration is as follows:At the heart of the device is the mesoporous oxide layer composed of a network of TiO 2nanoparticles that have been sintered together to establish electronic conduction.Typically,the ﬁlm thickness is ca.10µm and the nanoparticle size 10-30nm in diameter.The porosity is 50-60%.The mesoporous layer is deposited on a transparent conducting oxide (TCO)onaFigure 3.Incident photon to current conversion efﬁciency as a function of the wavelength for the standard ruthenium sensitizers N3(red line),the black dye N749(black curve),and the blank nanocrystalline TiO 2ﬁlm (blue curve).The chemical structures of the sensitizers are shown as insets.Reprinted with permission from ref 22.Copyright 2009American ChemicalSociety.Figure 4.Schematic overview of a dye-sensitized solar cell.Dye-Sensitized Solar Cells Chemical Reviews,2010,Vol.110,No.116601。

Silane_and_other_coupling_agent_coference

Improving the Corrosion Performance of Epoxy Coatings by Modiﬁcation with“Active”and“Non-active”Silane MonomersJi-Ming Hu∗,Wei-Gang Ji,Liang Liu,Jian-Qing Zhang and Chu-Nan Cao Department of Chemistry,Zhejiang University,Hangzhou310027,P.R.ChinaAbstractThe improvement of corrosion performance of epoxy coatings by modiﬁcation with“active”and“non-active”silane monomers is presented.“Active”silanes are deﬁned as ones having functional groups that can react with the epoxy group,usually amino-group silanes.“Non-active”silanes are not involved in a ring-opening reaction,but can be grafted onto the epoxy backbone by the condensation reaction between methoxy or ethoxy group in silanes and hydroxyl groups in the epoxy resin using an organic tin compound as the catalyst.FTIR spectroscopy results showed that both possible target reactions had successfully taken place between“active”/“non-active”silanes and epoxy resin.Electrochemical impedance spectroscopy(EIS)and accelerated corrosion tests(e.g.Machu test and boiling water test)show that the modiﬁed epoxy coatings provide better corrosion performance.KeywordsEpoxy resin,chemical modiﬁcation,silane agents,corrosion performance1.IntroductionIn recent years,a novel metal pretreatment based on the formation of silaneﬁlms has been developed to improve the corrosion resistance of metal substrate and the adhesion between painted organic coatings and metals[1–3].The silane agents used have a general structure(RO)3Si(CH2)n Y,where RO is a hydrolysable alkoxy group,such as methoxy(OCH3),ethoxy(OC2H5)or acetoxy(OCOCH3)and Y is an organofunctional group.The formation of silaneﬁlms is based on the condensa-tion reactions between silanols(Si–OH,hydrolysis product of alkoxy groups)and the metal hydroxyls(M–OH).The organofunctional silaneﬁlms deposited on the metal mainly act as a physical barrier against the permeation of water and corro-sive ions in the initial period[4–7].But once saturated with the electrolyte,the *To whom correspondence should be addressed.Tel.:86-571-87952318;Fax:86-571-87951895;e-mail: kejmhu@Silanes and Other Coupling Agents,V ol.5©Koninklijke Brill NV,Leiden,2009204J.-M.Hu et al.silaneﬁlms will lose the barrier effect and then the interfacial layer underneath plays a critical role in the subsequent corrosion inhibition process[3,8].A rapid upsurge of interest in silaneﬁlms has been found in theﬁeld of corrosion control of metals in recent years and investigations are being directed at developing this approach as an alternative to the currently-used carcinogenic chromating and pol-luting phosphating processes[3–6].However,single silaneﬁlms cannot effectively protect metals against corrosion due to their low thickness(only tens or a few hun-dreds of nanometers as reported in the literatures[9–12]).Therefore,in practice silaneﬁlms are usually applied in conjunction with organic coatings.Silanes can also be used to directly modify the paint systems(primers)to im-prove the corrosion performance of polymeric coatings.In our previous work[13], 3-glycidoxypropyltrimethoxysilane,GPTMS,was used as an additive and was sim-ply incorporated into epoxy coatings without pre-hydrolysis.The water absorption in the polymeric coatings was found to be reduced after the incorporation with GPTMS monomer.This may be due to continuous hydrolysis and condensation reactions of silane agent during the immersion test in aqueous corrosive solution, which enhanced the degree of crosslinking of the epoxy coatings.However,only a few silane monomers gave positive results by this simple direct incorporation. For example,some long-chain silanes(e.g.dodecyltrimethoxysilane,DTMS)de-teriorated the performance of the organic coatings,due to their poor compatibility with polymeric resins.Another problem is that because the silane monomers were simply mixed into organic coatings,the silane component may dissolve and sepa-rate out from the coatings during their application in an aqueous environment.To overcome these drawbacks,in this work we report a new methodology for chemical grafting two different groups of silane monomers,deﬁned as“active”and“non-active”silanes,onto the epoxy resin.The experimental results showed that the chemically modiﬁed epoxy coatings produced a signiﬁcant improvement in resis-tance against the penetration of water in polymeric resins and thus better corrosion protection of metals.As shown in Fig.1,two types of active groups exist in the epoxy backbone.One is the epoxide group,which can typically react with an amino-group resulting in ring opening.The other is a hydroxyl group,which is considered as the source of hydrophilicity in epoxy coatings.We deﬁne silanes containing an amino functionalFigure1.Chemical structure of epoxy resin.Improving the Corrosion Performance of Epoxy Coatings205(a)(b)Figure2.Possible reactions for grafting“active”(a)and“non-active”(b)silanes onto epoxy resin. group,e.g.γ-aminopropyltrimethoxysilane(γ-APS),as the“active”silanes due to their high reactivity towards the ring-opening reaction of epoxide group in the epoxy resin.Thus,the“active”silanes can be grafted onto the epoxy backbone, while other silanes,which do not contain amino group,are deﬁned as“non-active”silanes.However,by using these“non-active”silanes,the grafting reaction can still be achieved by the condensation reaction between the hydroxyl group in epoxy resin and alkoxy group in silanes.Possible reactions for there two kinds of grafting reactions are shown in Fig.2.2.Experimental2.1.Synthesis of Silane-Grafted Epoxy Resins2.1.1.Grafting of“Active”SilanesInto a500-ml three-neckedﬂask,equipped with a mechanical stirrer and a reﬂux condenser,were added a solution of20g(0.044mol)of epoxy resin(type E-44from Xuelian Resin Factory,Wujiang,China)in20ml of butanone.Different amounts(1, 3and5wt%with respect to epoxy resin)ofγ-APS were added to the solution.The mixture was allowed to react with stirring at70◦C for2h.The solvent(butanone) was removed by distillation and the silane-modiﬁed epoxy resin was obtained.206J.-M.Hu et al.2.1.2.Grafting of“Non-active”SilanesInto the above-mentioned container a solution of15g(0.033mol)of epoxy resin in 12ml of xylene and0.0128mol of various silanes(such amount is very close to but slightly lower than the mole amount of the hydroxyl group in E-44epoxy resin as derived from its hydroxyl value of0.10mol/100g)and certain amount of organotin compound as a catalyst were added.The mixture was allowed to react with stirring at90–100◦C for3h.The solvent(xylene)and the alcohol produced were removed under a reduced pressure and a viscous colourless product was obtained.Four silane monomers with different functional groups were used.Their chemical structures were given in Fig.3.2.2.Preparation of Coating Samples2024-T3aluminium alloys were used as the substrate.The curing agent was polyamide(type650from Yongzai Chemical Engineering,Pujiang,China).The mass ratio of epoxy resin to polyamide was10/8.The epoxy coatings were painted by using a spin coater(type KW-4A).All the coated samples were cured at35–45◦C for3days and then kept in a desiccator for at least2weeks before the tests.The thicknesses of the coatings obtained were35±2µm for“active”silane-modiﬁed coatings and45±2µm for“non-active”silane-modiﬁed coatings,as measured by induced eddy current technique.2.3.Electrochemical Impedance Spectroscopy(EIS)MeasurementsEIS was used to evaluate the water uptake of coatings.A three-electrode system was used,in which the coated sample acts as the working electrode,a saturated calomel electrode(SCE)as the reference,and a stainless steel disk as the counter electrode. The electrode surface area exposed to test solution(3.5wt%NaCl)was∼13.3cm2.Figure3.Chemical structures of GPTMS,DTMS,TEOS and VTES silane monomers.Improving the Corrosion Performance of Epoxy Coatings207 EIS measurements were performed on a VMP2multi-channel potentiostat(Prince-ton Applied Research,USA).A sinusoidal ac perturbation of20mV amplitude coupled with the open circuit potential was applied to the coating/metal system. The EIS test was performed in the frequency range of100kHz–100mHz.All the measurements were conducted at room temperature(∼25◦C).The experimental data were analyzed by using the commercial software EQUIVCRT developed by Boukamp[14].2.4.Glass Transition Temperature(T g)AnalysisDifferential scanning calorimetry(DSC)was employed to determine the T g of organic coatings before and after immersion in aqueous NaCl solution.The mea-surements were carried out on a Delta Series DSC7Thermal Analysis System under the protection of nitrogen atmosphere with a scan rate of20◦C/min in the temper-ature range from−30◦C to100◦C.The whole measuring process was carried out according to ASTM/D3418–82.2.5.FTIR SpectroscopyFTIR spectroscopy was used to characterize the structure of the silane-modiﬁed epoxy resin and to monitor the change in coating structure after immersion in salt solution.FTIR spectra were recorded on a Nicolet470spectrometer.Signals of 8repeated scans at a resolution of4cm−1were averaged before Fourier transfor-mation.The solution of silane-modiﬁed or pure epoxy resin was spread on a NaCl window and dried,and then a spectrum was taken.The spectra of cured epoxy coat-ings were obtained as follows:A small amount of the cured epoxy was ground to a ﬁne powder,mixed with KBr powder,and pressed into a pellet which was used to obtain the spectrum.2.6.Adhesion and Corrosion Performance Testing(Machu Test)The Machu test[15]was employed to evaluate the wet adhesion and corrosion performance of painted metals.The painted metal panels were cross-scribed by a sharp knife on the surfaces prior to the test,and then immersed into a solution of5wt%NaCl+0.6wt%H2O2at37◦C for1day.After24hours the solution was totally refreshed.After48hours of immersion,the panels were taken out and a tape was used to pull off the delaminated paint along the scribe lines.The photographs of the coated metals were taken by a digital camera after the testing.3.Results and Discussion3.1.“Active Silane”–Modiﬁed Epoxy CoatingsThe FTIR spectra of pure andγ-APS-modiﬁed epoxy resin are shown in Fig.4. It is evident that the peak intensity at910cm−1corresponding to epoxide group signiﬁcantly decreases after the silane modiﬁcation,indicating that the epoxide was chemically consumed by the silane agent.It is also clear that the peak intensity of208J.-M.Hu et al.Figure4.FTIR spectra of pure(1)and5%γ-APS-modiﬁed epoxy resins(2)(after[16]).–OH at3460cm−1increases due to the formation of new hydroxyl groups after the ring-opening reaction of epoxide group.In addition,the appearance of the peak at 1070cm−1corresponding to Si–O–C indicates that the silane component had been grafted onto epoxy resin.The FTIR results are in good agreement with the possible reactions as proposed in Fig.2.The quantitative evaluation of barrier properties of coatings themselves and their protectiveness against substrate corrosion was conducted by EIS measurements. By using a simple equivalent electrical circuit(EEC),containing the coating capac-itance(C c)and coating pore resistance(R po)presented in high-frequency domain, and the charge transfer resistance(R ct)and double layer capacitance(C dl)orig-inated by the metallic corrosion,as displayed in low-frequency region,EIS data were numericallyﬁtted and analyzed.More detailed explanation of EECs of epoxy-coated aluminum alloy systems can refer to our previous publications[17,18].The data analysis(see Fig.5)shows that after silane grafting,epoxy-coated Al alloy system has lower double layer capacitance(C dl)and higher charge transfer resistance(R ct),indicating better corrosion performance,except for5%silane mod-iﬁcation.As the indicators of adhesion and accelerated corrosion performance of organic-coated metals,the Machu test also shows that after silane modiﬁcation the protec-tive on efﬁciency of epoxy coating is signiﬁcantly improved.As shown in Fig.6, severe delamination is observed after testing on the pure epoxy coating/metal system.The polymer coating is almost completely detached from the metal sub-Improving the Corrosion Performance of Epoxy Coatings209Figure5.(a)C dl–and(b)R ct–time curves for pure andγ-APS-modiﬁed epoxy coatings/aluminium systems immersed in NaCl solution.Numbers marked are silane contents in modiﬁed epoxy coatings.Figure6.Photographs of pure(A)and1%(B),3%(C)and5%(D)γ-APS-modiﬁed epoxy coat-ing/aluminium systems after the Machu test.210J.-M.Hu et al.Table1.The T g values of pure and silane-modiﬁed epoxy coatings before and after immersion(after[16])Silane content in epoxy contents (wt%)T g(◦C) T g(◦C) Before immersion After immersion for100h066.9546.48−22.47 1.042.8946.203.31 3.044.5948.303.71 5.052.5935.59−17 strate(photo A),indicating that the adhesion between the coating and metal sub-strate was reduced signiﬁcantly after the testing.Moreover,the corrosion pits and corrosion products are visible on the substrate,indicating a serious corrosion of the metal substrate,i.e.,poor protection efﬁciency of pure epoxy coating.However,for all silane-modiﬁed coatings(photos B,C and D),no paint delamination has taken place after the testing.These results show that the adhesion and corrosion perfor-mance of epoxy coatings are obviously improved after the chemical modiﬁcation by silane monomers.The enhanced adhesion is believed to be related to the for-mation of a chemical Si–O–Me interfacial layer,as observed at silaneﬁlm-covered metals[19,20].The good consistency between the experimental results of EIS and the Machu test indicates a high reliability of both techniques in evaluating the per-formance of coated metals.The changes in the T g during immersion reﬂect the degree of polymer matrix plasticization and water/resin interactions[21].The T g value of the organic coating was usually found to decrease after immersion because the absorbed water mole-cules disrupt the inter-chain hydrogen bonds.Table1lists the T g values of pure and silane-modiﬁed epoxy coatings before and after immersion in aqueous NaCl solution.An exceptional decrease in the T g was found in pure epoxy coatings af-ter water uptake,as expected.However,the T g was found to slightly increase after immersion for1%and3%γ-APS-modiﬁed epoxy coatings.The increase in the T g value indicates that the degree of crosslinking of silane-modiﬁed coatings is improved after immersion,which ensures a better performance against water pene-tration.This indicates that in silane-modiﬁed coatings,the penetrated water brings about two competitive effects on coatings:one is the detrimental effect on coat-ings causing a decrease in the T g as mentioned above;the other is the beneﬁcial effect due to the reaction between water and silane components.This is because the cured coating system still contains hydrolysable alkoxy group–OCH3(see Fig.2a). The alkoxy group will hydrolyze and yield silanol(Si–OH)with the aid of water which can condense and form Si–O–Si structure.The hydrolysis of alkoxy group will consume the penetrated water resulting in a decrease in the amount of water uptake;on the other hand,the formation of Si–O–Si structure will increase the de-Improving the Corrosion Performance of Epoxy Coatings211Figure7.FTIR spectra of pure(a)and5%γ-APS-modiﬁed epoxy coatings(b)before(1)and after(2) immersion in NaCl solution for7days(after[16]).gree of crosslinking which will result in an increase in the T g and a decrease in water uptake.The competition between these two effects determines if the T g in-creases or not.The formation of Si–O–Si bond is supported by the FTIR spectra of pure and silane-modiﬁed epoxy coatings before and after immersion(Fig.7). The band at∼1030cm−1,typically attributed to substituted aromatic ring[22],is selected as reference assuming that the aromatic ring is stable during immersion. The absorption peak at∼1110cm−1is usually ascribed to phenyl ether[22],but such peak is also attributed to Si–O–Si in silaneﬁlms or in polysiloxanes[23]. The relative intensity of this peak is almost unchanged in pure epoxy coatings after 7days of immersion.However,for5%γ-APS-modiﬁed coating,the peak inten-sity increases noticeably after immersion(see Fig.7b).Therefore,the formation of Si–O–Si may be a reasonable explanation for the increased relative peak intensity at 1110cm−1during the immersion of silane-modiﬁed epoxy coatings.The obvious decrease of the T g in5%γ-APS-modiﬁed epoxy coating after immersion indicates that water mainly plays a negative(destructive)effect on crosslinking of polymeric backbone.The above experimental results measured in bulk polymeric coatings(changes in T g and FTIR spectra after1-week immersion shown in Table1and Fig.7) suggest that the crosslinking of“active”silane-modiﬁed epoxy coatings is basi-cally enhanced during their use in aqueous environment if proper content of silane monomer is used.But when excessive amount of silane is used the backbone of modiﬁed coatings is likely to be destroyed by the corrosive agents.In this sense,212J.-M.Hu et al.the water uptake of epoxy coatings is measured from the coating capacitance(C c), according to the Brasher–Kingsbury equation[24]:φw=log(C t/C0)log(εw),(1)whereφw is the volume fraction of water in organic coating,C t is the coating capac-itance at absorption time t,C0is the dry coating capacitance andεw is the dielectric constant of water,which is80.Theφw–time curves for pure andγ-APS-modiﬁed epoxy coatings during im-mersion in NaCl solution are shown in Fig.8.For1%γ-APS-modiﬁed coating, the amount of water uptake decreased signiﬁcantly as compared with unmodiﬁed pure epoxy coating.The reduction in water uptake may result from the consump-tion of water due to the hydrolysis of silane component,as well as the chemical modiﬁcation of polymeric matrix by hydrolysis products of silane agent during im-mersion.But for3%and5%γ-APS-modiﬁed epoxy coatings,φw–time curves do not show steady state behaviour but there is a continuous increase inφw during the whole immersion period.Theφw after a long-time immersion for these coatings is higher than that of pure epoxy coatings,indicating a deteriorating effect on the per-formance against water permeation by excessive addition of“active”silanes due to their high hydrophilicity.This result is consistent with T g and FTIR spectral results.Although the excessive amount of active silane monomers seems to result in a deteriorating effect on the backbone structure of the modiﬁed epoxy coatings dur-Figure8.φw–t1/2curves for pure andγ-APS-modiﬁed epoxy coatings immersed in NaCl solution. Numbers marked are silane contents in modiﬁed coatings(after[16]).Improving the Corrosion Performance of Epoxy Coatings213Figure9.FTIR spectra of(1)pure epoxy resin and(2)GPTMS-,(3)DTMS-,(4)TEOS-and(5)VTES-modiﬁed epoxy resins in different wavenumber regions;(a)2000–4000cm−1, (b)600–2000cm−1(after[17]).ing their use in a corrosive electrolyte,the real corrosion performance of modiﬁed coatings is still apparently better than unmodiﬁed coating.Accordingly,we suppose that the improved corrosion performance is more likely caused by the strengthening of coating/metal interface,besides the crosslinking effect of silanes.3.2.“Non-active”Silane-Modiﬁed Epoxy CoatingsIt should be noted that most of silanes cannot directly react with epoxy resin by ring-opening of epoxide group.But the methoxy or ethoxy group in silanes can be possibly condensed with hydroxyl group in epoxy resin,by which silane monomers are chemically grafted onto epoxy backbone.Figure9shows the FTIR spectra of pure and various“non-active”silanes-modiﬁed epoxy resins.The spectra of silane-modiﬁed epoxy resins are very similar to that of pure epoxy resin,but some differences can still be detected.In the high wavenumber region(2000–4000cm−1,shown in Fig.9a),it is evident that the peak intensity at about3500cm−1,corresponding to O–H stretching of the secondary hydroxyl in epoxy resin,decreases and even almost disappears after the silane modi-ﬁcation.This indicates that the hydroxyl groups were consumed by the modiﬁcation reaction,leading to a decrease in intensity of the absorption peak.In addition,as shown in the low wavenumber region(600–2000cm−1,shown in Fig.9b),after the silane modiﬁcation the peak intensity at1000–1100cm−1increases obviously due to the formation of Si–O–C bonds.These observations suggest that silane monomer had been successfully grafted onto epoxy resin.The possible reaction is depicted in Fig.2b.The formation of R–OH(methanol or ethanol),as proposed in the scheme,is consistent with the reﬂuxing phenomenon observed in the syn-thesis reaction.This phenomenon is expected to relate with the generation of some214J.-M.Hu et al.small molecular compounds considering that the reaction temperature is lower than the boiling point of xylene.Coating capacitance increase is used to investigate the sorption characteristics of water in organic coatings since this increase has been associated with water penetration into the coating,according to the Brasher–Kingsbury equation[24]as mentioned above.C c variations as a function of the immersion time in NaCl so-lution are shown in Fig.10a.For pure epoxy coating,after a long-time immersion the coating capacitance becomes steady,which means the coating attains a saturated state for water absorption.However,for silane-modiﬁed coatings,it is found that C c ﬁrst decreases after the initial stage and then reaches a steady value,in particular in the case of modiﬁcation with VTES and DTMS.More importantly,the C c values of silane-modiﬁed coatings are found to be signiﬁcantly lower than that of pure epoxy coating during the whole immersion time,indicating a signiﬁcant decrease in water uptake in silane-modiﬁed coatings.This may be caused by the hydrolysis reactions of active alkoxy groups in modiﬁed resins,by which the free state water absorbed is consumed.The condensation reactions between the hydrolyzed resins are also expected to improve the crosslinking of epoxy coatings.In addition,the hydrolysis products,methanol or ethanol,can also repel or restrict the permeation of water in the polymeric coatings as recently reported by Flis and Kanoza[25].The coating resistance(R c)is a measure of the porosity and degradation of the coating[26].Figure10b shows the R c–time curves for epoxy coating and silane-modiﬁed epoxy coatings during the immersion in NaCl solution.In the initial stage of immersion,the R c values of both the pure and silane-modiﬁed coatings decrease as a consequence of the development of conductive pathways in the coatings.For pure epoxy coating,after a long immersion time,R c almost stays steady except for a slight increase that may be caused by the accumulation of the corrosive products in conductive pathways.In the middle immersion stage(3–10days)the R c values of silane-modiﬁed coatings are found to be higher than that of pure epoxy coating,Figure10.(a)C c–t and(b)R c–t curves for pure(")and GPTMS(1),DTMS(P),TEOS(e)and VTES(E)-modiﬁed epoxy coatings during the immersion in NaCl solution(after[17]).Improving the Corrosion Performance of Epoxy Coatings215 indicating their better barrier performance against the penetration of electrolyte. In addition,for silane-modiﬁed coatings,R c shows aﬂuctuating trend,probably being caused by the hydrolysis and condensation of the active silane component in the coating matrix.The hydrolysis of silane component may reduce R c but the formation of Si–O–Si bonds due to the condensation of silanols may increase R c. Although,basically higher R c values are observed for silane-modiﬁed coatings,the lower values are still found in TEOS and DTMS-modiﬁed epoxy coatings after an intensive immersion.But the difference in R c between these modiﬁed and the unmodiﬁed epoxy coatings is insigniﬁcant.Figure11shows the C dl and R ct–time curves for pure and silane-modiﬁed epoxy/aluminum alloys during the immersion in NaCl solution.An increase in C dl can be seen for all coatings in the initial immersion stage(0–3days)showing the presence of the double layer under the coatings.For pure epoxy coatings,after a long-time immersion(5days),the C dl becomes steady which indicates the forma-tion of a stable interface.However,C dl shows aﬂuctuating trend for silane-modiﬁed coatings.This is the net result of both the delamination effect caused by corrosive attack and the healing effect by the formation of Si–O–Metal bonds during the im-mersion.The active alkoxy groups in modiﬁed coatings can be further hydrolyzed to form silanols(Si–OH),which may be further condensed onto metal surface by the formation of Si–O–Metal bonds,leading to the improvement in coating adhesion. All the silane-modiﬁed coatings demonstrate lower C dl than that of pure epoxy.For GPTMS-modiﬁed coatings,C dl decreases by approximately two orders of magni-tude.The lower C dl for silane-modiﬁed epoxy-coated system can be explained,on the one hand,by the better adhesion between the coating and metallic substrate as mentioned above and,on the other hand,by the obviously lower amount of water absorbed in the modiﬁed coatings as also mentioned above.Both of these result in a decrease of active area of metal/electrolyte interface.Figure11.(a)C dl–t and R ct–t(b)curves for pure(")and GPTMS(1),DTMS(P),TEOS(e)and VTES(E)-modiﬁed epoxy coatings during the immersion in NaCl solution(after[17]).216J.-M.Hu et al.All silane-modiﬁed samples have larger R ct values than the pure epoxy-coated sample.For the GPTMS-modiﬁed system,R ct value increases by approximately two orders of magnitude.The R ct values of other silane-modiﬁed samples increase by about one order of magnitude.This indicates that the silane-modiﬁed coatings have better anti-corrosion performance.The performance of silane-modiﬁed epoxy coatings was also evaluated by accel-erated Machu test.Figure12shows photographs of epoxy-coated aluminum alloys after the Machu test.As we have already observed from Fig.6A,severe delamina-tion has taken place after testing the pure epoxy coating/metal system,with some visible corrosion pits.The delamination is also observed for DTMS and VETS silane-modiﬁed coatings(photos B and D,respectively),but the areas of delami-nation are obviously smaller than that in case of pure epoxy coating.For GPTMS and TEOS silane-modiﬁed coatings(photos A and C,respectively),it is surprising toﬁnd that no paint delamination has taken place after the testing.These results show that the adhesion and corrosion performance of epoxy coatings are obviously improved after the chemical modiﬁcation with silane monomers,in particular for the cases of GPTMS and TEOS.The enhanced adhesion is believed to be related toFigure12.Photographs of(A)GPTMS-,(B)DTMS-,(C)TEOS-and(D)VTES-modiﬁed epoxy coating/metal systems after the Machu test(after[17]).Improving the Corrosion Performance of Epoxy Coatings217 the formation of a Si–O–Metal interfacial layer,as observed at silaneﬁlm-covered metals[19,20].Accordingly,the improved protectiveness of silane-modiﬁed epoxy coatings can be attributed to the reduction of water absorption in the coatings as well as to the enhanced adhesion with the substrates.4.Conclusions1.Theoretically,all alkoxysilanes can be chemically grafted onto epoxy resin,by the condensation reactions between alkoxy groups in silanes and hydroxyl group in epoxy,with the aid of oganotin catalyst.2.Certain“active”amino-silanes,e.g.γ-APS,can be easily grafted onto epoxyresin by the ring-opening reaction of epoxide group in the epoxy resin,with no need for a catalyst.3.The silane-modiﬁed epoxy coatings show better corrosion performance,due tohigher crosslinking of coatings,reduction of water uptake,and the strengthen-ing of coating/metal interface.AcknowledgementsThis work wasﬁnancially supported by the National Science Foundation of China (NSFC),contract no.50571090and the Zhejiang Provincial Natural Science Foun-dation,contract no.Y404295.Part of this work was funded by the State Key Laboratory for Corrosion and Protection.References1.D.Q.Zhu and W.J.van Ooij,anic Coatings49,42(2004).2.M.A.Petrunin,A.P.Nazarov and Yu.N.Mikhailovski,J.Electrochem.Soc.143,251(1996).3.D.Q.Zhu and W.J.van Ooij,Corrosion Sci.45,2177(2003).4.P.E.Hintze and L.M.Calle,Electrochim.Acta51,1761(2006).5.Y.C.Araujo,P.G.Toledo,V.Leon and H.Y.Gonzalez,J.Colloid Interface Sci.176,485(1995).6.A.Franquet,H.Terryn and J.Vereecken,Appl.Surface Sci.211,259(2003).7.H.Watson,P.J.Mikkola,J.G.Matisons and J.B.Rosenholm,Colloid Surfaces A161,183(2000).8.D.Q.Zhu and W.J.van Ooij,Electrochim.Acta49,1113(2004).9.A.Franquet,J.De Laet,T.Schram,H.Terryn,V.Subramanian,W.J.van Ooij and J.Vereecken,Thin Solid Films384,37(2001).10.A.Franquet,H.Terryn and J.Vereecken,Thin Solid Films441,76(2003).11.A.Franquet,C.Le Pen,H.Terryn and J.Vereecken,Electrochim.Acta48,1245(2003).12.A.Franquet,H.Terryn,P.Bertrand and J.Vereecken,Surface Interface Anal.34,25(2002).13.W.G.Ji,J.M.Hu,J.Q.Zhang and C.N.Cao,Corrosion Sci.48,3731(2006).14.B.A.Boukamp,Solid State Ionics31,18(1986).15.W.Machu,L.Schiffman and F.D.Archiv,Eisenhütenwesen37,679(1966).16.W.G.Ji,J.M.Hu,L.Liu,J.Q.Zhang and C.N.Cao,anic Coatings57,439(2006).17.W.G.Ji,J.M.Hu,L.Liu,J.Q.Zhang and C.N.Cao,Surface Coatings Technol.201,4789(2007).18.J.T.Zhang,J.M.Hu,J.Q.Zhang and C.N.Cao,anic Coatings51,145–151(2004).。

水性环氧富锌底漆的制备与性能研究

lm Vol.54 N o.l Jan. 2021水性环氧富锌底漆的制备与性能研究李时珍“2,王德修2,李新雄:，罗小虎13,朱日龙、谢佳武2，李季\吴大旺3，刘娅莉^(1.湖南大学化学化工学院，湖南长沙410082; 2.邦弗特新材料股份有限公司，湖南宁乡410600;3.黔南民族师范学院化学化工学院，贵州都匀558000)[摘要]为开发新型水性富辞底漆，在开发具有优良水可稀释性能的水性固化剂基础上，研究了一种环境友好的双组分水性环氧富锌底漆。

将辞粉等颜填料和各种助剂均勾分散于以乙醇为溶剂的水性环氧树脂固化剂中，制备出了稳定的固化剂组分；通过溶剂析出率测试和高速离心测试，对辞粉在水性环氧树脂固化剂中的稳定性进行了研究；通过对涂膜的防腐蚀性能的测试，研究了辞粉含量、粒径、形状对涂膜性能的影响；通过扫描电子显微镜 (SEM)分析和热重分析（TGA)对制得的水性环氧富锌底漆进行了表观形态分析和耐热性研究。

结果表明：选择粒径为3.80 |xm左右的球形和椭球形混合辞粉，添加量为75%~80%时制得的水性环氧富锌底漆力学性能最好，具有最佳的防腐蚀效果。

[关键词]水性环氧富辞底漆；环氧固化剂；锌粉；耐蚀性能[中图分类号]TQ630.1 [文献标识码]A[文章编号]1001-1560(2021 )(H-0104-08Preparation and Properties of Water-borne Zinc-rich Epoxy Primer LI S h i-z h e n g12,W A N G D e-x i u2,LI X i n-xiong2,L U O X i a o-h u13,Z H U R i-l o n g1,X I E J i a-w u2,LI Ji',W U D a-w a n g3,L I U Y a-l i1(1. D e partment of Chemistry a n d C h e mical Engineering, H u n a n University,C h a n g s h a 410082,C h i n a；2. Banfert N e w Materials Co., Ltd., Ningxiang 410600,C h i n a；3. School of Chemistry a n d C h e m i c a l Engineering, Q i a n n a n No r m a l University for Nationalities, D u y u n 558000,China)A b s t r a c t：B a s e d o n the development of water - borne curing agent with excellent water dilubility,a n environment -friendly two - c o m p o n e n t water-borne zinc-rich epoxy primer w a s studied. Firstly, zinc p o w d e r a n d the various additives were evenly dispersed in the water-based epoxy resin curing agent with ethanol as solvent to prepare stable curing agent. T h e stability of zinc p o w d e r in water-borne epoxy curing agent w a s studied b y solvent precipitation rate test a n d h i gh-speed centrifugation test. T h e effect of zinc p o w d e r content, particle size a n d shape on the film performance w a s studied by testing the anti-corrosion performance of the film. T h e water-borne epoxy zinc rich primer w a s characterized by scanning electron microscopy (S E M) a n d thermogravimetric analysis (T G A)to study the morphology a n d heat resistance. T h e results s h o w e d that the mechanical a n d anticorrosive properties of waterborne epoxy zinc rich primer are the best w h e n the particle size of spherical and ellipsoidal m i x e d zinc p o w d e r is about 3.8 (xm a n d the content of zinc p o w d e r is 75%~80%.K e y W o r d s：water-borne zinc - rich epoxy pr i m e r；epoxy curing agent；zinc p o w d e r；anticorrosion performance〇前言富锌涂料具有电化学的阴极保护作用和物理屏蔽双重耐蚀性能[〃]，是重防腐蚀涂料的主要品种之一。

光电子技术与器件光显示与显示器件

cr ystals and conjugated polymer composites[刊,中]/刘艳山(华南理工大学高分子光电材料与器件研究所,特种功能材料教育部重点实验室.广东,广州(510640)),王藜//高等学校化学学报.―2007,28(3).―596599采用有机金属液相法制备了平均粒径为5nm的CdSe纳米微球(ns CdSe),并将其与共轭聚合物(MEH PP V或P3H T)共混制备了太阳电池器件。

透射电镜(TEM)、紫外可见吸收光谱(UV Vis)及荧光光谱(PL)研究结果表明,CdSe纳米晶呈均匀的球状颗粒,在近红外区具有良好的吸收和荧光性能;加入CdSe纳米晶能够有效地淬灭共轭聚合物的荧光。

给出了在AM1.5模拟太阳光(光强为100mW/cm2)照射下,ns CdSe/M EH PP V共混体系太阳电池器件性能的测试结果。

图3参16(严寒)TM914.42007043656基于ZnO光阳极的染料敏化太阳电池=Development and application of ZnO based photo anode in dye sensitized so lar cells[刊,中]/盛显良(中科院化学研究所有机固体重点实验室.北京(100080)),赵勇//化学进展.―2007,19 (1).―5965全面介绍了基于ZnO光阳极的染料敏化太阳电池的研究和应用现状,特别是ZnO光阳极的制备方法,包括传统的手术刀法、丝网印刷技术和电沉积自组装方法,以及最近发展起来的机械挤压法、化学液相沉积法、化学汽相沉积法和低温水热法等,对不同制备薄膜方法的工艺条件和优缺点进行了综述。

同时介绍了微/纳米复合结构和直线电子传输对光电转换效率提高的作用。

图8表1参44 (严寒)TM914.42007043657工作状态对MIS/IL太阳电池反型层电荷的影响=Effect of output voltage on charge in inversion layer in MIS/IL solar cell[刊,中]/丁扣宝(浙江大学信息与电子工程学系.浙江,杭州(310027))//太阳能学报.―2007,28(1).―2831在一定输出电流及光照条件下,数值求解半导体器件基本方程,获得输出电压和反型层电荷面密度等参数。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

Performance of ZnO dye-sensitized solar cells with various nanostructures as anodesZhifeng Liu *,Yabin Li,Chengcheng Liu,Jing Ya,Wei Zhao,Lei E,Dan Zhao,Li AnDepartment of Materials Science and Engineering,Tianjin Institute of Urban Construction,300384Tianjin,PR Chinaa r t i c l e i n f oArticle history:Received 13December 2010Received in revised form 14March 2011Accepted 11April 2011Available online 20April 2011Keywords:ZnOPhotoanode Solar cell Ef ﬁciency NanotubesPorous nanosheetsa b s t r a c tSeveral types of ZnO nanostructured ﬁlms,including nanoparticle ﬁlms,ordered porous ﬁlms,nanorod ﬁlms,nanotube ﬁlms and porous nanosheet ﬁlms,were fabricated and used as photoanodes in dye-sensitized solar cells (DSSCs).Their performance was described and compared in detail.DSSCs based on ordered porous ﬁlms presented relatively higher conversion ef ﬁciency (0.7%)than that of DSSCs based on nanoparticle ﬁlms (0.2%).DSSCs based on nanorod ﬁlms showed the maximal ef ﬁciency up to 1.3%,mainly attributed to the c -axis oriented structure which could provide the faster conduction pathway for charge transport.Although nanotube ﬁlms with large surface area could absorb more dyes,the ef ﬁciency of DSSCs based on nanotube ﬁlms was only 1.2%and wasn ’t as high as that of nanorod DSSCs,which may be caused by the charge-carrier recombination losses at the defects formed in the etching.Moreover,the porous ZnO nanosheets photoanode showed an improved ef ﬁciency by 2.5times as compared to ZnO nanosheets samples due to increase the dye loading and light harvesting.Ó2011Elsevier Masson SAS.All rights reserved.1.IntroductionOver the last few years,the dye-sensitized solar cells (DSSCs)have been attracting lots of attention since the initial report by Gratzel ’s group due to their potential application for development of low-cost,large area photovoltaics [1].DSSCs with TiO 2nano-porous ﬁlms have been demonstrated with a power conversion ef ﬁciency of 11%in which a thick layer of nanoporous ﬁlm provides a large surface area of anchoring the light harvesting dye molecules [1,2].However,slow electron percolation through the inter-connected nanoparticles and the charges recombination between injected electrons and electron acceptors in the electrolyte hinder the performance of TiO 2DSSCs [3,4].In order to increase the ef ﬁ-ciency of energy conversion,many semiconductor ﬁlms,such as ZnO [5],SnO 2[6]and Nb 2O 5[7],since their band gap energy are similar to TiO 2,were proposed to improve charge transfer.ZnO is a II e VI compound n-type semiconductor with unique properties such as transparency in the visible and high infrared re ﬂectivity,acoustic characteristics,high electrochemical stability and excellently electronic properties,which could be an alternative material for solar cell.To achieve high performance of DSSCs,ZnO photoanode is required to possess high surface area as well as goodelectrical,electrochemical and structural properties.However,ef ﬁciency of DSSCs based on ZnO nanoparticle ﬁlms was only reached 0.4%early in 1994by Gratzel ’s group due to trap-limited diffusion for electron transport,a slow mechanism that may limit device ef ﬁciency,especially at longer wavelengths [2,4,8].Then a version of ZnO nanorods/nanowires DSSCs had been introduced and got good score [9e 11].In addition,2D ZnO nanostructures,such as nanosheets and nanobelts,have also been studied for DSSC applications on account of the fact that they also have a special structure,which can improve the conversion ef ﬁciency [12,13].Although there are many reports on the preparation and applica-tion of ZnO photoanodes in DSSCs via different techniques [12e 17],comparison among different ZnO photoanodes ranging from nanoparticles,nanorods,nanotubes and nanosheets is important,which can give some revelations about the effect of ZnO photo-anode structure on the DSSCs cell performance.Here we report and discuss the performance of ZnO DSSCs with nanoparticle ﬁlms,ordered porous ﬁlms,nanorod ﬁlms,nanotube ﬁlms and porous nanosheets ﬁlms,fabricated by different techniques.2.Experimental 2.1.Substrates treatmentAll the ZnO photoanode ﬁlms of DSSCs in our experiment were prepared on the indium tin oxide (ITO,10U cm -1)glass substrates,*Corresponding author.Tel.:þ862223085236;fax:þ862223085110.E-mail address:tjulzf@ (Z.Liu).Contents lists available at ScienceDirectSolid State Sciencesjournal ho mep age:www.elsevier.co m/locate/ssscie1293-2558/$e see front matter Ó2011Elsevier Masson SAS.All rights reserved.doi:10.1016/j.solidstatesciences.2011.04.005Solid State Sciences 13(2011)1354e 1359which were ultrasonically rinsed for15min in acetone,iso-propyl alcohol and ethano absolution,respectively.2.2.Preparation of ZnO nanoparticleﬁlmsZnO powders were synthesized by mixing0.1M zinc nitrate water solution with0.1M diethanolamine,adding a small quantity of cetyltrimethylammonium bromide(CTAB)n-butanol solution. The mixed solution was then allowed to age in a thermostat at80 C for24h.The product was washed with deionized water,centri-fuged at a high speed andﬁnally dried.As-prepared ZnO powders were mixed with ethanol and stirred overnight,resulted in a colloidal suspension with ZnO content of20wt%.At last,the suspension was coated onto the ITO glass substrate by scalpel using scotch tape as frame and spacer.After drying in air,the nanoparticle ﬁlms were calcined in air up to500 C at a heating rate of2 C/min.2.3.Preparation of ordered ZnO porousﬁlmsOrdered ZnO porousﬁlms were prepared by cathodic electro-deposition with polystyrene spheres templates as pore-forming assistant[18].2.4.Preparation of ZnO nanorod arraysLarge-scale arrays of oriented single-crystal ZnO nanorods was successfully fabricated on electrodeposited ZnO seed layer by aqueous solution method from zinc nitrate and hexamethylene-tetramine at low temperature(typically95 C)[19].2.5.Preparation of ZnO nanotube arraysZnO nanotube arrays were obtained by chemical etching as-prepared nanorod arrays using0.1M alkaline solution(KOH)at 85 C for1h,which were thoroughly rinsed in distilled water in order to remove any residual salt or amino complex,and then were dried in air at room temperature.2.6.Preparation of porous ZnO nanosheetsPorous ZnO naosheetsﬁlms were prepared by electrochemical deposition atﬁxed water bath temperature(70 C)in an aqueous solution containing0.5M Zn(NO3)2solution with pH¼5.0Æ0.1 using polystyrene spheres(PS)template as pore-forming assistant [20].The ZnO nanosheets were also prepared by electrochemical deposition method under the same conditions only without PS template assistant.2.7.Assembly of ZnO dye-sensitized solar cellsThe above several types of ZnOﬁlms were used as photoanodes in DSSCs with2-m m thickness,which can be controlled by technologic parameters[18e21],sensitized in a0.05mM ethanol solution of Ruthenium(II)cis-di(thiocyano)bis(2,20-bypyridyl-4,4’dicarboxylic acid)(N3)dyes for at least12h at60 C.The excess unanchored dyes were rinsed off using absolute ethanol and dried in air,then covered with platinum sheet as counter electrodes.The internal space of the cell wasﬁlled with liquid electrolyte(0.5M LiI,0.05M I2)dissolved in acetonitrile by capillary action.2.8.CharacterizationMorphology of theﬁlms was observed by PHILIPS XL-30envi-ronment scanning electron microscopy(ESEM)and transmission electron microscopy(TEM,JEOL100CX-II).X-ray diffraction(XRD) patterns of theﬁlms were recorded with a Rigaku D/max-2500 using Cu K a radiation(l¼0.154059nm).Optical transmittance of photoanodeﬁlms was examined by DU-8B UV/VISdouble-beamFig.1.Typical SEM images of different ZnOﬁlms used as photoanodes in DSSCs(a)nanoparticleﬁlms(the inset is a TEM image);(b)ordered porousﬁlms;(c)nanorod arrays;(d) nanotube arrays.Z.Liu et al./Solid State Sciences13(2011)1354e13591355spectrophotometer.Photocurrent of the ZnO DSSCs was measured under irradiation of a xenon lamp (80mW cm À2)with global AM1.5condition,and photocurrent e voltage curves of the ZnO DSSCs were obtained using a potentiostat (TD3691,Tianjin Zhonghuan CO.,LTD,China).It should be noted that in order to lessen the errors on the ef ﬁciency measurements,ﬁve cell samples based on every type ZnO ﬁlms were used.3.Results and discussionFig.1shows the typical SEM images of ZnO nanoparticle ﬁlms,ordered porous ﬁlms,nanorod arrays and nanotube arrays,respectively.The nanoparticle ﬁlms are uniform and composed of particles with the diameter about 10e 20nm,illuminated as the inset one in Fig.1(a).In Fig.1(b),porous ZnO ﬁlms prepared by electrodeposition method on ITO glass substrates covered with PS array templates show an ordered,honeycomb-like macroporous structure.The average diameter of pores is about 350nm.The ﬁlm is so robust that no cracking or deformation occurs during the removal of PS by immersing in toluene.Well-aligned ZnO nanorod arrays in large-scale can be successfully fabricated on electro-deposited ZnO seed layer (Fig.1(c)),and the well-de ﬁnedcrystallographic planes of the hexagonal nanorods can be clearly identi ﬁed from the high-magni ﬁcation SEM image (inset image in Fig.1(c)),indicating the nanorods grow along the [001]direction.Fig.1(d)illustrates the representative SEM image of ZnO nanotube arrays transformed from as-prepared ZnO nanorod arrays by etching in an alkaline solution.Interestingly,the etching is selective and only takes place in the center of nanorods,which is attributed to the two polar planes,(001)and (001),in wurtzite ZnO crystal.The two planes with high surface energy are metastable,while the nonpolar planes parallel to c -axis are the most stable planes with a lower surface energy.As a result,the etching rate of the polar (001)plane is faster than that of the nonpolar planes [22].Moreover,ZnO as an amphoteric oxide can react with hydroxyl ions and produce soluble salts as the following:ZnO þ2OH À%ZnO 2À2þH 2O(1)Therefore,ZnO nanotubes are formed by selective etching on the polar and nonpolar planes in case there is an appropriate OH Àconcentration in the solutions.Fig.2presents the ordered ZnO porous ﬁlms before and after sensitized by N3dyes.As expected,the sensitized ﬁlm (Fig.2(b))appears brown due to anchoring N 3dyes.Other photoanodes settled with dyes,including nanoparticle/nanorod/nanotube ﬁlms,appear similar to ordered porous ﬁlms,revealing that the dyes are well adsorbed on the surface of ZnO ﬁlms.Fig.3gives the photocurrent e voltage curves of the ZnO DSSCs with nanoparticle ﬁlms,ordered porous ﬁlms,nanorod ﬁlms and nanotube ﬁlms,while their photoelectrochemical properties are listed in Table 1.During the photocurrent measurements,the cell ef ﬁciency (h )is expressed by the following equation:h ¼ðV oc J sc FF Þ=P in(2)FF ¼V opt J opt =V oc J sc(3)Fig.3.Photocurrent-voltage curves of DSSCs with different ZnO photoanode ﬁlms.Table .1Parameters of ZnO dye-sensitized solar cells with different photoanode ﬁlms.Photo anode ﬁlms V oc (mV)J sc (mA cm À2)FF(%)h (%)Nano particle ﬁlms 3000.92580.2Ordered porosu ﬁlms 468 1.96630.7Nanorods ﬁlms 442 3.7562 1.3Nanotubes ﬁlms3864.70531.2Fig.4.XRD pattern of the ZnO nanorodarrays.parison of ordered ZnO porous ﬁlms before and after sensitized by N3dyes.Z.Liu et al./Solid State Sciences 13(2011)1354e 13591356where P in is the power of incident white light,FF is ﬁll factor,V opt and J opt are voltage and current for maximum powder output,and V oc and J sc are open circuit photovoltage and short circuit photo-current,respectively.Comparing in terms of open circuit voltage,short circuit photocurrent,ﬁll factor and power conversion ef ﬁciencies,it can be found that morphology of ZnO photoanode ﬁlms have profound impact on the performances of DSSCs.DSSCs based on ZnO nanoparticle ﬁlms (nanoparticle DSSCs)composed of inter-connected spherical particles show the lowest ef ﬁciency (0.2%)due to charge-carrier recombination losses at grain boundaries between nanoparticles [2].The average values of open circuit voltage (V oc )and short circuit current (J sc )for nanoparticle DSSCs are 300mV and 0.92mA cm -2,respectively.While the values are 468mV and 1.96mA cm -2for DSSCs based on ZnO ordered porous ﬁlm.And the DSSCs based on ZnO ordered porous ﬁlm perform higher conversion ef ﬁciency (0.7%)than nanoparticle DSSCs (0.2%),mainly caused by the fewer defects in ordered porous ﬁlms than in nanoparticle ﬁlms.Moreover,the photo-electric conver-sion ef ﬁciency of DSSCs based on ZnO porous ﬁlms will be improved by decreasing the pore diameter because of the increasing surface area for dye adsorption from the porous structure.The higher conversion ef ﬁciency (1.3%)for DSSCs based on ZnO nanorod ﬁlms (nanorod DSSCs)can be ascribed to its special nanostructure.Fig.4exhibits the typical XRD pattern of ZnO nanorods arrays ﬁlm.The result shows all the diffraction peaks can be indexed to the wurtzite structure.In comparison with standard powder diffraction pattern (PDF#65e 3411),much stronger intensity of (002)peak to others manifests nanorods are well preferentially oriented in the direction of c -axis,which coincides with the SEM results.Well-aligned single-crystal nanorods can provide faster electron transportation channels than grain boundaries which associate with traps or/and barriers.And larger surface area of nanorod arrays will bene ﬁt the dyes adsorbing and improve the light harvest ef ﬁciency.Furthermore,injected electrons can transport directly through oriented nano-rods to conducting substrates,which greatly reduces the recom-bination losses.Fig.5gives the optical transmittance spectra of the ZnO nanorod and nanotube arrays attached by N3dyes,both of the ﬁlms show the maximum adsorption peak around 535nm cor-responding to N3dyes.Although the stronger and wider absorp-tion peak for nanotube ﬁlms indicates that more dyes are adsorbed and then result in the increase of J sc ,DSSCs based on nanotube ﬁlms (nanotube DSSCs)present an ef ﬁciency of 1.2%,slight lower than that of nanorod DSSCs.In addition,J sc and V oc of nanotube DSSCs are 4.70mA cm -2and 386mV,while the values of nanorod DSSCs are 3.75mA cm -2and 442mV.As mentioned above,the higher J sc of nanotube DSSCs is attributed to the larger surface area that anchors more dyes.More defects may be produced during the process of ZnO nanotubes by chemical etching,leading to the electrolyte direct contact with the ITO substrate and the lower V oc .Both surface area and defects simultaneously play crucial roles on the performance of nanotube DSSCs.Actually,the conversion ef ﬁciency of nanotube DSSCs isn ’t as high as one of nanorod DSSCs,due to the charge-carrier recombination losses at the defects of the direct contact between electrolyte and ITO substrate,even though the nanotube arrays with large surface area can absorb more dyes.The ZnO nanosheets can be obtained in an aqueous solution containing 0.5M Zn(NO 3)2solution with pH ¼5.0Æ0.1whentheFig.5.Optical transmittance spectra of ZnO nanorod/nanotube ﬁlms.Fig.6.SEM images of ZnO nanosheets at different conditions (a)potential is À0.8V and without PS template;(b)potential is À0.8V and with PS template;(c)potential is À0.9V and with PStemplate.Fig.7.Photocurrent-voltage curves of the DSSCs based on ZnO nanosheets.Z.Liu et al./Solid State Sciences 13(2011)1354e 13591357potential is À0.8V or À0.9V [20].The morphologies of the ZnO nanosheets obtained were examined by SEM analysis.Fig.1shows the typical SEM images of ZnO nanosheets with different morphologies.These ZnO nanosheets standing on the ITO substrate exhibit the regular hexagon end planes with about 4-5um in diameter and 100nm in thickness.Moreover,it can be seen from Fig.6(a)that the surface of ZnO nanosheets electrodeposited under above conditions without PS template assistant is smooth,and there is a mass of gaps existing among the crystals.However,ZnO nanosheets (as shown in Fig.6(b)and (c))with porous structure can be synthesized under the same conditions using PS template as pore-forming assistant.During the preparation of porous ZnO nanosheets,the ITO substrate covered with PS was ﬁrstly soaked in Zn(NO 3)2aqueous solution at 70 C for 30min,which could weaken the fastness of PS particles in some extent before the electrochemical deposition reaction [17].Then,at the action of electric ﬁeld,the chemical reaction taking place at the working electrode (the cathode in this case)is as the following:Zn ðNO 3Þ2þ2e À/ZnO Y þNO À3þNO À2(4)The growth of ZnO at the spaces of PS further weakened the fastness of PS particles and the weak PS,which was caused both by the soaking in electrolyte and the hustle of ZnO growth,might drift toward the solution due to the action of electric ﬁeld.It was known that Zn precursor in solution had positive charge and PS had negative charge,so there was an electrostatic attraction between the Zn precursor and PS.As a result,PS were allured and embedded by ZnO crystals during the electrodeposition.Porous ZnO nanosheets ﬁlms could be obtained after the samples were soaked in toluene to dissolve the PS.It should be noted that the electrodeposition potential has an important role on the morphology of porous ZnO nanosheets.In Fig.6(b),PS can be only embedded in the bottom of ZnO nanosheets (porous ZnO nano-sheets(B))at À0.8V because the drift speed of PS is slower than the growth of ZnO crystals.However,the drift speed of PS increases with the relative increasing of potential.So PS can be embedded in whole ZnO nanosheets (porous ZnO nanosheets(W))at À0.9V and uniform porous structure can be obtained after soaking the sample in toluene to dissolve the PS (as shown in Fig.6(c)).Fig.7gives the photocurrent e voltage curves of the DSSCs based on the ZnO nanosheets using N3as sensitizer,and the photoelectrochemical properties of these DSSCs listed in Table 2The cell performances of the DSSCs based on porous ZnO nano-sheets(B)and porous ZnO nanosheets(W)are higher than ZnO nanosheets based DSSCs and are expected to be high.The average values of open circuit voltage (V oc )and short circuit current (J sc )for porous ZnO nanosheets(B)based DSSCs are 348mV and 3.3mA cm -2,respectively,While the values are 307mV and 2.1mA cm -2for DSSCs based on ZnO nanosheets.Moreover,the porous ZnO nanosheets(B)based DSSCs exhibit an improved ef ﬁ-ciency by 100%corresponding to that of ZnO nanosheets.And the porous ZnO nanosheets(W)based DSSC reached a total ef ﬁciency of 1.0%,which is 2.5times as high as that of the nanosheets based DSSC (0.4%).All of these results indicate that,porous ZnO nano-sheets can enhance cell performance compared to the nanosheets without porous structure.As we all know,except the effect of recombination of the photo-excited electrons on the cell properties,one reason for the low conversion ef ﬁciency of traditional DSSCs is the dye loading on semiconductor.Another reason may be explained by the solar light utilization.The ZnO nanosheets photoanode with porous structure can overcome the two limitations above.The dye loading and solar light harvesting of ZnO nanosheets based DSSCs can be schematically illustrated in Fig.8,where the porous structure may be formed by using PS template as pore-forming assistant during the growth of ZnO crystals;the dye is very easy to load on the porous ZnO nanosheets photoanodic ﬁlm due to the enlargement of surface area,which signi ﬁcantly increases the cell perfor-mances.From Fig.7and Table 2,compared to the ZnO nanosheets,the porous ZnO nanosheets,both porous ZnO nanosheets(B)and porous ZnO nanosheets(W),show an improved cell performances.Especially,there is an overall increase of cell performance for porous ZnO nanosheets(W)based DSSCs due to the uniform pore distributing in the whole nanosheets.It should be noted that the enhanced solar light harvesting behavior is also associated with the suf ﬁcient dye loading.Besides,the nanosheets with porous structure are very ef ﬁcient on increasing the light harvesting,which signi ﬁcantly improve the current and is bene ﬁcial for the ﬁnal conversion ef ﬁciency.Moreover,as shown in Fig.8(c),some solar light can further re ﬂect after through the pore in the nano-sheets,resulting in maximum light harvesting in porous nano-sheets photoanode.Table 2Parameters of DSSCs with different ZnO nanosheets photoanode ﬁlms.Photoanode ﬁlm V a (mV)J sc (mA cm À2)FF(%)h (%)ZnO nanosheets 307 2.1550.4ZnO nanosheets(B)348 3.3540.8ZnO nanosheets(W)3734.2531.0Fig.8.The schematic diagram of the dye loading and light harvesting for ZnO nanosheets based DSSC.Z.Liu et al./Solid State Sciences 13(2011)1354e 135913584.ConclusionsSeveral types of ZnO nanostructuredﬁlms,including nano-particleﬁlms,ordered porousﬁlms,nanorodﬁlms,nanotubeﬁlms and nanosheetﬁlms,were fabricated and used as photoanodes in dye-sensitized solar cells.Their performance was described and compared in detail.DSSCs based on ordered porousﬁlms presented relatively higher conversion efﬁciency(0.7%)than that of DSSCs based on nanoparticleﬁlms(0.2%),due to the charge-carrier recombination losses at grain boundaries for nanoparticleﬁlms. DSSCs based on nanorodﬁlms showed the maximal efﬁciency up to 1.3%,mainly attributed to the c-axis oriented structure which could provide the faster conduction pathway for charge transport. Although nanotubeﬁlms with large surface area could absorb more dyes,the efﬁciency of DSSCs based on nanotubeﬁlms was only1.2% and wasn’t as high as that of nanorod DSSCs,which may be caused by the charge-carrier recombination losses at the defects formed in the etching.In addition,the porous ZnO nanosheets photoanode showed remarkable inﬂuence on theﬁnal performance of the DSSCs.I e V characteristic measurement indicates an improved efﬁciency by2.5times as compared to ZnO nanosheets samples due to increase the dye loading and light harvesting. AcknowledgementsThe authors gratefully acknowledgeﬁnancial support from the Key Project of Chinese Ministry of Education(No.208008),China Postdoctoral Science Foundation Funded Project(No.20080440674) and China Postdoctoral Science Special Foundation(No.201003294). References[1] B.O’Regan,M.Grätzel,Nature353(1991)737.[2]M.Grätzel,Prog.Photovolt.Res.Appl.8(2000)171.[3]N.Kopidakis,K.D.Benkstein,J.van de Lagemaat,A.J.Frank,J.Phys.Chem.B.107(1999)11307.[4]N.Saito,H.Haneda,T.Sekiguchi,N.Ohashi,I.Sakaguchi,K.Koumoto,Adv.Mater.14(2002)418.[5]S.Karuppuchamy,T.Yoshida,T.Sugiura,Solid State Ionics151(2002)19.[6]N.G.Park,M.G.Kang,K.M.Kim,K.S.Ryu,S.H.Chang,Langmuir20(2004)4246.[7]P.Guo,M.A.Aegenter,Thin Solid Film351(1999)290.[8]w,L.E.Greene,J.C.Johnson,R.Saykally,P.D.Yang,Nat.Mater.4(2005)455.[9]M.Guo,P.Diao,X.Wang,S.Cai,J.Solid State Chem.178(2005)3210.[10]J.Tornow,K.Schwarzburg,J.Phys.Chem.C111(2007)8692.[11]H.Gao,G.Fang,M.Wang,N.Liu,L.Yuan,C.Li,L.Ai,J.Zhang,C.Zhou,S.Wu,X.Zhao,Mater.Res.Bull.43(2008)3345.[12] E.Hosono,S.Fujihara,I.Honma,H.S.Zhou,Adv.Mater.17(2005)2091.[13] C.Lin,H.Lin,J.Li,X.Li,J.Alloys Compd.462(2008)175.[14]Y.Sun,G.M.Fuge,M.N.R.Ashfold,Chem.Phys.Lett.396(2004)21.[15]Z.B.Fang,Y.Y.Wang,X.P.Peng,X.Q.Liu,C.M.Zhen,Mater.Lett.57(2003)4187.[16]M.J.Zhang,L.D.Zhang,G.H.Li,W.Z.Shen,Chem.Phys.Lett.363(2002)123.[17]endre,D.S.Boyle,P.O’Brien,D.Binks,D.West,D.Coleman,Adv.Mater.14(2002)1221.[18]Z.Liu,Z.Jin,J.Qiu,X.Liu,W.Wu,W.Li,Semicond.Sci.Technol.21(2006)60.[19]Z.Liu,L.E,J.Ya,Y.Xin,Appl.Surf.Sci.255(2009)6415.[20]Z.Liu,Z.Jin,W.Li,X.Liu,J.Qiu,W.Wu,Mater.Lett.60(2006)810.[21]Z.Liu,C.Liu,J.Ya,L.E,Solid State Sci.12(2010)111.[22]G.W.She,X.H.Zhang,W.S.Shi,X.Fan,J.C.Chang,C.S.Lee,S.T.Lee,C.H.Liu,Appl.Phys.Lett.92(2008)053111.Z.Liu et al./Solid State Sciences13(2011)1354e13591359。