Electrochemical Deposition of Conformal Semiconductor Layers in Nanoporous Oxides for Sensitized Photoelectrodes.

Nanoporous photoelectrodes with photoactive semiconductors have been investigated for various energy applications such as solar cells and photoelectrochemical cells, but the deposition of the semiconducting materials on the nanoporous electrodes has been very challenging due to pore clogging or complete pore filling. Here, we propose a band alignment model that explains the morphology of the electrochemically deposited semiconductor layer on the semiconducting nanoporous oxide electrode. Briefly, the coating material with a conduction band edge higher (i.e., more negative) than that of the electrode material forms a conformal coating, which maintains the initial nanoporous structure. As a result, a conformal CdSe layer can be electrodeposited onto TiO2 nanotubes, which can be used as a photoelectrode of a sensitized solar cell. The electron dynamics studies revealed that the CdSe-sensitized TiO2 nanotube electrode exhibited faster charge transport and slower charge recombination than its dye-sensitized counterpart, which has been ascribed to the passivation of surface traps and the physically blocked back-electron transfer by the CdSe layer as well as the higher conduction band of CdSe.

Introduction

Sensitized solar cells (SSCs) with various sensitizers including molecular dyes,[1,2] quantum dots,[3−5] and thin inorganic layers[6] have attracted much attention as potential candidates for next-generation low-cost modestly efficient solar cell devices. Among the sensitizers, the thin inorganic layers possess advantages of prominent light absorption capability,[7] tunable electronic properties,[8] and direct deposition on the oxide electrodes over other sensitizers.[9] A typical thin inorganic layer SSC consists of an electron-transporting phase, a sensitizer, and a hole-transporting phase. The electron-transporting phase is typically a nanoporous oxide semiconductor where the pores are generally less than 100 nm for obtaining larger surface area. The sensitizer covers the surface of the electron-transporting nanostructure. The liquid or solid hole-transporting phase is infiltrated into the pores of the sensitized electron-transporting electrode. Therefore, it is critical to maintain the nanoporous structure after depositing the sensitizer to obtain a typical SSC structure.

The electrochemical deposition has been one of the most popular processes to fabricate inorganic thin films[10,11] and to fill the pores of electrically insulating nanostructures like porous aluminum oxide films.[12] However, it has been challenging to control the morphology of deposited materials within nanoporous semiconducting electrodes (e.g., TiO2 electron-transporting electrodes of SSCs) owing to the small pore size and electrical conductivity of the electrode materials. For instance, it has been shown that the semiconductors usually do not fill the pores of the electrodes but form an overlayer on top of the electrodes owing to the predominant deposition at the entrance of the pores.[13] One approach to overcome this clogging issue has been demonstrated by filling the pores with the deposition electrolyte followed by the electrochemical deposition in an inert electrolyte.[13] We previously developed a general electrochemical bottom-up growth approach by adjusting the ambipolar diffusion length (characteristic reaction length) by controlling the ionic strength of the electrolyte sovlent[14] where the bottom-up growth of p-type CuInSe2 (CIS) in ordered nanoporous n-type TiO2 electrodes was demonstrated. In the same study, we also showed that CIS completely filled the entire TiO2 pore structure. However, for the traditional SSC structure, a conformal coating of the inorganic sensitizer is required to ensure the open pore structure. A couple of papers[15,16] reported the electrochemical deposition of the conformal CdSe or CdTe layer on the surface of ZnO nanowire arrays having average diameters/pores of hundred nanometers where the pores are interconnected to each other. On the other hand, there has been no report on the electrochemical deposition of conformal semiconductor layers on the electrically conducting nanoporous network with isolated pores less than 100 nm. Furthermore, no mechanistic study has been reported regarding the factors determining the deposition morphology between the complete pore filling and the conformal coating within a nanoporous semiconductor network.

In this study, we propose a band alignment model that determines whether a semiconductor would fill the pores of the semiconducting oxide electrode completely or it would form a conformal coating on the surface of the electrode by electrochemical deposition. As a proof-of-concept example of this model, we discuss the results of CdSe deposition on the TiO2 nanotube (NT) arrays, and its growth behavior was compared with that of the CIS/TiO2 NT counterpart. The interfacial charge transfer between the host nanoporous electrode and the coating material is found to be affected by the relative position of their conduction band edge, which in turn determines the deposition morphology (i.e., conformal coating vs pore filling). For example, we found that the conduction band offset between TiO2 and as-deposited CdSe results in the conformal coating of CdSe on TiO2 NT arrays. Also, the electron transport and recombination properties of the CdSe-sensitized solar cell (CdSe-SSC) consisting of the conformal CdSe-coated TiO2 NT arrays were compared with those of its conventional dye-sensitized counterpart (DSSC). At a fixed photoelectron density, the CdSe-SSC with a cobalt-based electrolyte system exhibit 10 times faster transport and 2 times slower recombination than the DSSC with the same electrolyte.

Experimental Section

TiO2 Nanotube Preparation

Oriented TiO2 NT arrays were prepared by electrochemically anodizing Ti foil (Alfa, 0.25 mm, 99.5% purity) in a two-electrode cell with a Pt counter electrode. The anodization electrolyte consisted of 0.15 M ammonium fluoride (Aldrich, 99.9%) and 3.5 wt % water in formamide (Aldrich, 99.9%). The Ti foils were biased at 20 V for 2 h at room temperature, resulting in 2 μm thick amorphous NT arrays. As-prepared NT films were rinsed with water and ethanol followed by the subsequent thermal crystallization at 400 °C in air for 1 h.

CdSe Deposition

The electrochemical deposition was carried out using a computer-controlled potentiostat (EG&G, PAR283) with a three-electrode cell with a Pt mesh counter electrode and a Ag/AgCl quasi-reference electrode (calibrated against [Fe(CN)6]3–/4–). The CdSe was deposited on the TiO2 NT electrode by a potential pulse method (−0.85 V vs Ag/AgCl for 0.5 s and open circuit for 2.5 s) with a deposition solution composed of 20 mM CdCl2 (Aldrich, 99.99%) and 4 mM H2SeO3 (Aldrich, 99.999%) in absolute ethanol (Pharmco-Aaper, 200 prf). After deposition, CdSe was crystallized by annealing at 400 °C for 1 h (ramp rate of 5 K·min–1) under a N2 atmosphere.

Thin Absorber Solar Cell

CdSe-coated TiO2 NT films were assembled into sensitized solar cells according to the procedure described in our previous paper.[17] Briefly, the device was assembled in a sandwich configuration with the nanotube film facing a counter electrode, prepared by spreading a droplet of 5 mM H2PtCl6 in 2-isopropanol onto the commercial F:SnO2 (FTO, TEC15) conducting glass substrate and subsequently firing it at 350 °C. In order to prepare the electrolyte solution, 0.2 M [Co(phen)3](ClO4)2 (tris[1,10-phenanthroline]cobalt(II) perchlorate), 0.02 M Co3+ complex ([Co(dtb-bpy)3](ClO4)3, tris[4,4′-di-tert-butyl-2,2′-bipyridine]cobalt(III) perchlorate),[18] and 0.2 M LiClO4 were dissolved in the mixture solvent composed of acetonitrile and ethylene carbonate (4:6 in volume ratio). The Co3+ complex was used instead of the chemical oxidizing agents like NOBF4 (nitrosodium tetrafluoroborate) to prevent the possible production of nitric oxide (NO) species.[19] The active area after cell assembly was 0.5 cm2.

Characterization

The crystalline structure and microstructure of the CdSe-coated TiO2 NT films were characterized by X-ray diffraction (XRD) and field-emission scanning electron microscopy (FE-SEM). Transport and recombination properties were measured by intensity-modulated photocurrent spectroscopy and intensity-modulated photovoltage spectroscopy as described previously.[20] For these measurements, the solar cells were probed with a modulated beam of 680 nm light superimposed on a relatively large background (bias) illumination, also at 680 nm. The probe and bias light entered the cell from the counter electrode side.

Results and Discussion

As described in the Introduction, our previous study revealed that the CIS completely fills the pores of TiO2 NT arrays during the electrochemical deposition.[14] As a result, the sample forms a very nice p–n bulk heterojunction structure (Figure a). However, the conduction band edge position of CIS (+0.2 eV vs NHE)[21] is lower than that of TiO2 (−0.5 eV vs NHE),[22] hindering efficient injection of photoelectrons from CIS to TiO2, which is not ideal for a p–n bulk heterojunction solar cell. So, we selected CdSe as an alternative to CIS because its conduction band edge position (−0.8 eV vs NHE)[23] is known to be higher than that of TiO2, and thus, the photoelectrons can be easily injected to TiO2.[24] One could expect a similar p–n bulk heterojunction structure with CdSe if the same electrochemical deposition technique were used. However, as shown in Figure b, the morphology of CdSe after the electrochemical deposition was totally different from that of CIS. CdSe forms a conformal coating on the nanoporous TiO2 NT arrays, resulting in a porous structure that is suitable for a conventional three-phase sensitized solar cell. Comparing the NT morphology before and after the deposition (i.e., upper left image in Figure a and lower left image in Figure b, NT walls get thicker after CdSe deposition, but pores are still open throughout the NT. The analysis of SEM images with higher magnifications (not shown) revealed that the thickness of the NT walls increases from 10–15 to 20–25 nm, whereas the inner diameter of the NT pores decreases from 60–70 to 50–60 nm. This result indicates that the conformal CdSe layer with a thickness of ∼5 nm covers both inner and outer surfaces of the NTs. So, what mechanism during the electrochemical deposition allows CdSe to form a conformal layer as opposed to the complete pore filling by CIS? Given the reason why we chose CdSe as opposed to CIS (i.e., conduction and edge positions), we hypothesized that the band alignment between the host n-type electrode and guest p-type semiconductor materials would affect the behavior of the electrochemical deposition.

Figure 1

(a) SEM images of CIS-deposited TiO2 NT arrays, showing a complete bottom-to-up pore filling behavior (upper left: before deposition; lower left: during the deposition (2.9 h); right: overlayer formation after a 3.8 h deposition at −0.95 V vs Ag/AgCl). Note that panel (a) was adopted from our previous publication.[14] (b) SEM images of CdSe-coated TiO2 NT arrays where CdSe was coated conformally on the surface of TiO2 NTs after 43200 pulse cycles (i.e., 6 h of deposition at −0.85 V vs Ag/AgCl).

Figure shows a schematic illustration of the effects of the in situ band alignment between the electrode and as-deposited coating materials on the final deposition morphology. During the cathodic electrochemical deposition using a negative bias potential, the electrons needed for the deposition (i.e., the reduction of the reactants) should be supplied from the conducting substrate to the electrode/electrolyte interface. It is noteworthy that all the deposition mechanisms in this study were discussed assuming that the solvent with a low ionic strength like the absolute ethanol was used for the deposition. With such a solvent, the ambipolar diffusion length of the electrons in the nanoporous TiO2 electrode filled with the electrolyte is sufficiently short, resulting in the bottom-up growth.[14] In the case where the conduction band edge of the coating material is lower (i.e., more positive) than that of the electrode material (Figure a), the electron transfer from the electrode to the coating layer is facilitated due to the appropriate band alignment. Therefore, the deposition takes place at both the coating material/electrolyte interface (routes 1 and 2 in panel (a)) and the nearby electrode/electrolyte interface (route 3 in panel (a)), resulting in the complete filling of the pores with the coating material. On the other hand, if the conduction band position of the coating material is higher (i.e., more negative) than that of the electrode material, the electron transfer from the electrode to the coating layer is obstructed by the inappropriate band alignment. In this case, deposition will stop once the electrode surface is covered by the coating layer with a certain thickness (routes 2 and 3 in panel (a)), causing electrons to migrate further through the electrode to find a suitable reaction site (i.e., the electrode/electrolyte interface; route 1 in panel (a)). Consequently, the coating layer will grow conformally on the electrode surface from bottom to top. This model can also explain the conformal coating of CdSe on the nanoporous ZnO film reported by other groups.[15,25] Indeed, ZnO and TiO2 are known to have very similar conduction band levels, as well as valence band positions and band gap values,[26] which results in similar band alignments with CdSe.

Figure 2

Schematic bottom-up deposition mechanisms for (a) complete pore filling and (b) conformal coating.

Figure a shows the XRD patterns of a CdSe-coated TiO2 NT film before and after annealing at 400 °C under a N2 atmosphere. The as-deposited CdSe is amorphous, as no peak other than crystalline anatase peaks (denoted as A) from the TiO2 NTs and Ti metal peaks (denoted as *) can be observed. The XRD pattern of the TiO2 NT electrode (i.e., the as-deposited sample) is consistent with previous reports on anodized TiO2 NT systems.[27,28] After annealing, the as-deposited amorphous CdSe layer was crystallized to the hexagonal wurtzite structure with characteristic three peaks between 23 and 28° (with a CuKα radiation) denoted as C. Because the as-deposited CdSe layer is amorphous, the in situ band alignment between CdSe and TiO2 that was assumed based on the conduction band position of crystalline CdSe has to be confirmed. Figure b compares the photoresponse of the TiO2 NT electrodes in the presence and in the absence of the as-deposited CdSe, from which one can determine the in situ band alignment between CdSe and TiO2 during the deposition. A three-electrode electrochemical setup similar to the electrochemical deposition was used with an aqueous polysulfide electrolyte (0.1 M Na2S, 0.1 M S, and 0.1 M NaOH). A blue-light-emitting diode with a characteristic wavelength of 470 (±10) nm (i.e., 2.64 eV) and an optical power of 5 mW was used as a visible-light source. Given the estimated band gap energy of each material (i.e., 1.72 eV for CdSe and 3.2 eV for TiO2), it was assumed that most incident photons would be absorbed by CdSe. For better comparison, the lowest current during the pulse illumination was set to zero. It is evident from the difference between the two current density profiles that the as-deposited CdSe is photoactive and the excited photoelectrons are being injected to the TiO2 conduction band as illustrated in the inset of Figure b. If the conduction band edge of the as-deposited CdSe were lower than that of TiO2, no photocurrent should have been observed from this measurement. The direction of the photocurrent (i.e., anodic photocurrent), which was confirmed by comparing to that of a normal dye-sensitized solar cell, also supports the band alignment assumption between as-deposited CdSe and TiO2.

Figure 3

(a) X-ray diffraction patterns of CdSe-coated TiO2 NT arrays before and after annealing at 400 °C for 1 h under a N2 atmosphere where C, A, and the asterisk (*) indicate peaks of CdSe (wurtzite, JPCDS #77-2307), TiO2 (anatase, JPCDS #21-1272), and Ti metal substrate (JPCDS #44-1294), respectively. (b) Photocurrent density profiles of TiO2 NT electrodes with and without the as-deposited CdSe layer under pulse illumination by 470 nm light where the inset shows an illustration of the band alignment and the resulting charge injection between the as-deposited CdSe layer and the TiO2 NT.

Figure shows the transport (τc) and recombination (τr) time constants for a CdSe-sensitized TiO2 NT solar cell (CdSe-SSC) and a dye-sensitized counterpart (DSSC), both of which use a cobalt-based electrolyte system. Both τc and τr exhibit the usual power-law dependency on photoelectron density and Voc, respectively. The power-law dependency of τc can be explained by a model in which electrons perform an exclusive random walk between trap sites that have a power-law distribution of waiting (release) times in the form of t–1−α, where parameter α is in the range from 0 to 1 and can be related to the shape of the trap distribution.[29,30] From the best fits of the τc data in Figure to the expression τc (∝D–1) ∝ n1–1/α,[31] the respective α values for the CdSe-SSC and the DSSC were determined to be 0.29 and 0.27, respectively. Similar α values are understandable, as the two solar cells consist of the same electron-transporting phase (i.e., the TiO2 NT arrays). At a given photoelectron density or Voc (i.e., the same quasi-Fermi level), the CdSe-SSC exhibits 10 times faster transport and 2 times slower recombination than the DSSC, which should lead to a much higher charge-collection efficiency (ηcc). The slower recombination of the CdSe-SSC than the DSSC can be attributed to the existence of the CdSe conformal layer, which can physically/energetically prevent the electron recombination from TiO2 to the oxidized electrolyte species. The faster electron transport in the CdSe-SSC is also likely associated with the changes of the surface properties. Passivation of surface traps by the conformal CdSe layer could lead to faster electron transport, owing to the reduced number of traps that the electrons visit and thus the reduced time that the electrons spend before they are collected by the Ti foil.[32] In addition, the conduction band offset at the CdSe/TiO2 interface could confine the injected electrons in the region relatively far from the TiO2 surface, which should similarly result in faster transport.

Figure 4

Transport time constants (τc, solid symbols) as a function of photoelectron density and recombination time constants (τr, open symbols) as a function of open circuit photovoltage (Voc) for CdSe-sensitized (circles) and N719-dye-sensitized (squares) TiO2 NT solar cells where the solid/dotted lines show the best fit of the data.

Conclusions

In summary, we have proposed a band alignment model that determines the morphology of the electrochemically deposited semiconductor layer on the semiconducting nanoporous oxide electrode. The electron microscopy combined with a photoelectrochemical measurement revealed that the semiconductor forms conformal coating on the electrode surface when its conduction band is higher than that of the electrode (e.g., CdSe/TiO2 system). On the other hand, the semiconductor fills the pores of the electrode completely in the case of the opposite band offset. The amorphous as-deposited CdSe conformal layer crystallized to the hexagonal wurtzite phase after annealing at 400 °C for 1 h under a N2 atmosphere. The CdSe-sensitized TiO2 NT electrode exhibited faster charge transport and slower charge recombination than the dye-sensitized counterpart. The band alignment model proposed in the study provides a convenient avenue of determining materials combination for the morphology control during the electrochemical deposition.