Impact of Mg Doping on Performances of CuGaO 2 Based p‑Type Dye-Sensitized Solar Cells Ade ̀ le Renaud, † Laurent Cario,* ,† Philippe Deniard, † Eric Gautron, † Xavier Rocquefelte, † Yann Pellegrin, ‡ Errol Blart, ‡ Fabrice Odobel, ‡ and Ste ́ phane Jobic* ,† † Institut des Mate ́ riaux Jean Rouxel, Universite ́ de Nantes, CNRS, 2 rue de la Houssinie ̀ re, BP 32229, 44322 Nantes, Cedex 03, France ‡ Universite ́ LUNAM, Universite ́ de Nantes, CNRS, Chimie et Interdisciplinarite ́ : Synthè se, Analyse, Mode ́ lisation (CEISAM), UMR 6230, 2 rue de la Houssinie ̀ re, 44322 Nantes, Cedex 03, France ABSTRACT: p-Type dye solar cells (p-DSSCs) have been receiving much attention due to their potential role in the elaboration of future tandem dye solar cells with a photoanode and a photocathode built upon n-type and a p-type semiconductors, respectively. So far, NiO appears as the most widely used semiconductor in p-DSSCs. Yet this material suffers from several drawbacks, e.g., a low electrical conductivity and a low redox potential that limit the photovoltaic performances. In that framework, delafossite compounds may be regarded as appropriate substitutes of nickel oxide in relation to their intrinsic optoelectronic properties. Here we report on the nanostructuration of CuGaO 2 and its Mg doped derivatives via hydrothermal conditions with Pluronic P123 as surfactant. It appears that a low amount of magnesium helps in preparing samples with higher specific surface areas and stimulates enhanced conversion efficiencies. Beyond a given Ga/Mg substitution rate, a new Cu 1‑x (Ga,Mg)O 2 phase is formed that contains a large amount of Cu vacancies and of structural defects. This new phase shows lower photovoltaic performances compared to those for slightly doped derivatives which suggests that Cu vacancies and/or structural defects limit the mean free path of holes within the photocathode. ■ INTRODUCTION Solar cells based on the sensitization of a p-type semiconductor are of growing interest to researchers. 1-7 The development of such devices appears as the mandatory step to achieve tandem dye-sensitized solar cells (t-DSSCs) with a photoanode built upon a conventional n-type semiconductor (n-SC, e.g., TiO 2 and ZnO) 8,9 and a photocathode built upon a p-type semiconductor (p-SC, e.g., NiO and CuO), 1,2,10 both electro- des being sensitized by two distinct dyes. The fabrication of such devices would open up the door to higher efficiencies in the conversion of sunlight into electrical power compared to n- DSSCs alone as demonstrated by He et al., 11 Odobel et al., 12 and Nattestad et al. 13 Unfortunately, p-DSSCs performances are currently very low compared with those of n-DSSCs, and intensive work has to be dedicated to the quest to find new nanostructurated p-SCs to favor higher open-circuit voltage (V OC ), short-circuit current (J SC ), and fill factor (ff). 14 To date, the most frequently used p-SC in p-DSSCs remains by far NiO with a common conversion efficiency (η) of a few tens of a percent. Nevertheless, it is worth noticing that a real breakthrough was recently discovered by Powar et al. 15 where NiO is coupled with the outstanding PMI-6T-TPA dye as a sensitizer and tris(1,2-diaminoethane)cobalt(II/III) complexes as a redox mediator to reach a η value of 1.3%. Moreower, new p-SCs were recently explored to replace NiO and significant success were achieved with delafossites. 3-6,16,17 The limited performances of NiO-based solar cells were so far in part attributed to a too high energy position of the top of its valence band. In that framework, we have embarked recently on the synthesis of p-type CuGaO 2 -delafossite nanoparticles in ethylene glycol (EG) mediated hydrothermal conditions. 18 Namely electrochemical impedance spectroscopy (EIS) meas- urements on NiO and CuGaO 2 samples evidenced an increase of the flat band potentials (FBP) by almost 200 mV shifting from the binary to the ternary compounds. Unfortunately, this significant enhancement of the FBP does not go along with an increase of the p-DSSCs performances, owing to a still low short circuit photocurrent (Jsc) likely due to unoptimized p- SC/dye and dye/mediator interfaces favoring counter-produc- tive reactions. This may also be associated with a relatively low absorbance of the photocathode suggesting that the size of prepared CuGaO 2 particles is still too large leading to low specific surface area and finally a low amount of adsorbed dyes on its surface. An increase of J SC in CuGaO 2 based DSSCs appears therefore as the key point to outperform NiO based cells. For that purpose, it is crucial to obtain smaller Received: July 22, 2013 Revised: December 10, 2013 Published: December 11, 2013 Article pubs.acs.org/JPCC © 2013 American Chemical Society 54 dx.doi.org/10.1021/jp407233k | J. Phys. Chem. C 2014, 118, 54-59