Applied Surface Science 351 (2015) 840–845 Contents lists available at ScienceDirect Applied Surface Science journal h om epa ge: www.elsevier.com/locate/apsusc Mechanism of O 3 sensing on Cu 2 O(1 1 1) surface: First principle calculations Hela Ouali a,b , Caroline Lambert-Mauriat a,∗ , Laurent Raymond a , Ahmed Labidi b a Aix-Marseille Université, Université de Toulon, CNRS, IM2NP UMR 7334, 13397 Marseille Cedex 20, France b Université de Carthage, Laboratoire des Matériaux, Molécules et Applications – IPEST, BP 51, La Marsa, 2070 Tunis, Tunisia a r t i c l e i n f o Article history: Received 10 May 2015 Accepted 4 June 2015 Available online 12 June 2015 Keywords: Cu2O surface Ozone detection Ab initio calculation Gas sensors a b s t r a c t Combining experiments and first-principle calculations, mechanism of ozone detection on Cu 2 O(1 1 1) surface is proposed. Results show that O 3 is adsorbed without dissociation with E ads = -2.89 eV and is completely dissociated in O 2 plus oxygen adsorbed with total enthalpy H = -3.01 eV. These two reac- tions leads to a shift in the level Fermi of about 0.45 eV, which induces a p-doping. Thus the reaction of dissociation of O 3 on Cu 2 O(1 1 1) surface leads to a decrease in electrical resistance, as experimentally observed. © 2015 Elsevier B.V. All rights reserved. 1. Introduction Since many years, metal oxide semiconductors are used in gas sensor applications. If first devices were mainly based on Sn 2 O sen- sitive layer [1], nowadays more oxides are used like WO 3 [2,3], ZnO [4], CuO [5,6], etc. Despite their wide success bases on their low cost, their good responses to several gases and their easy imple- mentation in integrated systems, these sensors sin by their lack of selectivity. Indeed these devices are based on the variation of the electrical resistivity of the metal-oxide in the presence of gas, which depends on both the oxido-reductor behavior of the target gas and the doping of the semiconductor. To improve the selectivity of these gas sensors our team has made the choice to develop multi-sensor devices. The principle is to associate several sensors with different sensitive layers and to analyze the different responses by numerical post-treatment such as neuronal network [7]. In particular we have studied WO 3 , n-type semiconductor to detect O 3 , NO x and CO [8,9]. In the continuity of these works we first have been interested in CuO p-type semiconductor as layer for O 3 detection [10]. We found that O 3 adsorbed without dissociation is responsible of the decrease observed in the electrical resistance of the sensor device, by enhancement of the p-doping of the layer. An other p-type semiconductor cuprous oxide, Cu 2 O, appears as a promising candidate in such sensor devices for H 2 S [11] or ethanol vapor [12]. Recently we have thermally synthesized pure Cu 2 O ∗ Corresponding author. Tel.: +33 0491288974. E-mail address: caroline.mauriat@im2np.fr (C. Lambert-Mauriat). successfully [13] and first response to ozone are presented here. But a better understanding of the gas adsorption mechanism is needed, in particular at the atomic scale. Therefore numerical sim- ulations based on the density functional theory (DFT) are carried out in order to study the interaction of Cu 2 O with O 3 . The crystal- lographic structure of Cu 2 O is cubic (space group O 4 h = Pn3m) and contains two Cu 2 O per cell, each oxygen atom being four-fold coor- dinated by copper atoms (Fig. 1). The most stable surface is found to be the non polar (1 × 1) reconstructed (1 1 1) one by DFT + U calcula- tions [14]. Thus, we present computational results of the adsorption of O 3 molecules on the Cu 2 O(1 1 1) surface and correlate them to experimental response under ozone. The manuscript is organized in the following way. After expos- ing the computational details, accompanied by partial results on both the bulk and the surface (Section 2), experiments settings are briefly presented (Section 3). Then the dissociative and non disso- ciative adsorption of ozone is presented (Section 4) before detailing the detection mechanism in Section 5. Finally we give a short con- clusion of this work. 2. Computational details Ab initio calculations are carried out using the siesta code [15,16]. The exchange–correlation potential is treated within the generalized gradient approximation (GGA) using the Perdew–Burke–Ernzerhof functional [17]. In all calculations core electrons are treated within the frozen core approximation using norm-conserving Troullier–Martins [18] pseudopotentials. In addi- tion, nonlinear core corrections are included in pseudopotentials http://dx.doi.org/10.1016/j.apsusc.2015.06.017 0169-4332/© 2015 Elsevier B.V. All rights reserved.