Visualization of nanocrystalline CuO in the grain boundaries of Cu2O thin films and effect on band bending and film resistivity

Direct evidence for the presence of a CuO structure in the grain boundaries of Cu2O thin films by chemical vapor deposition is provided by high resolution automated phase and orientation mapping (ASTAR), which was not detectable by classical transmission electron microscopy techniques. Conductive atomic force microscopy (CAFM) revealed that the CuO causes a local loss of current rectification at the Schottky barrier between the CAFM tip and Cu2O. The suppression of CuO formation at the Cu2O grain boundaries is identified as the key strategy for future device optimization.

The p-type oxide semiconductor Cu 2 O has a band gap of 2.09 eV and a wide range of carrier concentrations is reported in the literature (10 9 -10 16 cm −3 ). [1][2][3][4][5] These characteristics allow different kinds of applications such as photovoltaics, photocatalysis, thin-film transistors, or resistive switching devices. [6][7][8] However, the material is only metastable at room temperature and ambient pressure and shows a tendency to form precipitates of secondary phases. [9][10][11] The formation of secondary phases is typically controlled by process parameters, which influence the thermodynamics and kinetics of the deposition. 12,13 In addition, it is still debated whether the observation of Cu(ii) on the surface of airexposed Cu 2 O films by XPS 11,[14][15][16] corresponds to the presence of a CuO native layer. For maximum conversion efficiencies of photovoltaic devices, the formation of CuO at the interface between Cu 2 O and the n-type layer needs to be avoided. 14 The presence of secondary phases has a tremendous effect on the transport properties of the material. The grain boundaries in polycrystalline Cu 2 O are known to be more conductive than the grain interior. 17 In thin films, oxidation of the copper ions to Cu(ii) at low film thicknesses was observed by X-ray photoelectron spectroscopy. 18 Since CuO has a band gap of 1.5 eV and a carrier concentration of 10 19 cm −3 , 19,20 the increased conductivity in the grain boundaries was ascribed to the presence of Cu(ii). 18 However, no evidence for the existence of a CuO structure was provided.
In the present study, high resolution transmission electron microscopy (TEM) based automatic phase mapping was used to identify the crystal structure at the grain boundaries of Cu 2 O thin films deposited by aerosol assisted metal organic chemical vapor deposition (AA-MOCVD) with a pointto-point resolution of 0.25 nm. Conductive atomic force microscopy (CAFM) was used to investigate the local current transport characteristics of the polycrystalline Cu 2 O films. a Electronic mail: j.deuermeier@campus.fct.unl.pt b Electronic mail: david.munoz-rojas@grenoble-inp.fr 2166-532X/2018/6(9)/096103/6 6, 096103-1 © Author(s) 2018 The Cu 2 O thin films were deposited in a homemade aerosol assisted metal organic chemical vapor deposition (AA-MOCVD) system. Copper(ii) trifluoroacetylacetonate (98%, Strem Chemical) was used as the precursor and ethanol as the solvent. A 0.01 M solution was prepared by mixing 1.85 g of the precursor in 0.5 l of solvent, following with 10 min magnetic bar stirring. The depositions were conducted for 220 min at 335 • C with additional samples prepared at 305 • C and 350 • C. During the deposition, the consumption rate of the precursor solution was maintained at around 1.5 ml/min. Gas flows were 2 l/min N 2 (carrier gas 1) and 5 l/min air with 3 l/min N 2 (carrier gas 2). The pressure in the deposition chamber was close to atmospheric pressure. Different substrates were used: samples for TEM/ASTAR characterization were deposited on silicon whereas CAFM measurements were conducted on films deposited on indium-tin oxide (ITO) coated glass. Corning glass was used for X-ray diffraction and Raman measurements reported in the supplementary material.
The morphology and structural properties of Cu 2 O thin films were investigated by transmission electron microscopy with a JEOL 2010 LaB6 microscope operating at 200 kV with a 0.19 nm pointto-point resolution. In addition, a JEOL 2100F FEG TEM equipped with the precession assisted automated crystal phase and orientation mapping tool ASTAR from NanoMEGAS was used. 21,22 An Asylum Research MFP-3D atomic force microscope was used to obtain topographies, local current-voltage characteristics, and current maps. The latter were measured in contact mode with the voltage applied to the chuck while the tip (Nanoworld CONTPt) remained at ground potential. In order to evaluate the electrical properties of the Cu 2 O/ITO interface with an Agilent 4155C semiconductor parameter analyzer, Ohmic top contacts of 5 nm Ni and 60 nm Au were deposited through shadow masks by electron beam evaporation.
TEM results of a deposited Cu 2 O film are presented in Figs. 1(a) and 1(b). A cross section image of the Cu 2 O thin film is shown in Fig. 1(a), in which columnar-like grains are oriented perpendicular to the substrate. The length of the grains can reach almost 1 µm. Figure 1 pattern, the Cu 2 O structure was observed, showing a preferential orientation with (111) planes parallel to the substrate surface. By comparing the obtained diffraction pattern with the CuO reference [as shown in Fig. 1(b)], no diffraction reflections associated with the CuO structure could be found in the deposited Cu 2 O film by SAED. Other commonly used techniques such as X-ray diffraction and Raman spectroscopy were also used and no CuO was detected (see the supplementary material).
The cross section of the deposited Cu 2 O film was characterized by the ASTAR technique. The precession angle used in the experiments was 1.16 • . The diffraction patterns were obtained with a step size of 1 nm. The obtained phase maps are presented in Figs. 1(c)-1(e). Three phases were identifiedmarked with red, blue, and green-corresponding to Cu 2 O, CuO, and silicon, respectively. Note that no CuO can be detected on the surface of the films. From the cross section, columnar Cu 2 O grains with a width of about 400 nm and a length of 1 µm are observed, similar to the TEM results shown in Fig. 1(a). The closer view of the phase mapping in Fig. 1(d) clearly shows that the blue colored CuO phase is mostly located at the grain boundaries. In particular, the amount of CuO is significantly larger near the interface between the substrate and the Cu 2 O film, where the Cu 2 O grain size is smaller. The existence of the CuO phase in the grain boundaries is in line with the results obtained with XPS, which showed the presence of Cu(ii). 18 The reconstructed correlation index map is presented in Fig. 1(e). The gray scale map was obtained by plotting the value of the correlation index for the succession of the electron diffraction patterns at each location. It indicates that the obtained phase mapping data have rather good reliability.
The CAFM results are shown in double sweeps between 0 V and +5 V, followed by 0 V and −5 V, was calculated. A general accordance between grain boundary regions in the topographies and the areas of high conductance is observed. Note that an increased current level in the grain boundary regions could be an artifact related to the increased contact area in the grain boundary region due to the surface topography. However, the barrier under negative substrate bias is only observed in the grain region. A CAFM study reported in the literature revealed a Schottky barrier between the cantilever tip and the surface of doped silicon, which disappeared when the sample was illuminated by the laser of the AFM with a photon energy above the band gap of silicon (1.85 eV). 23 In the case of our samples, conducting the current-voltage measurement of the grain boundary region in the dark (without the AFM laser emitting with an energy of 1.44 eV nor ambient light) showed no influence on the current rectification. Hence, photo-induced carriers are excluded as a reason for the absence of a barrier between the AFM tip and the grain boundaries. In order to investigate the influence of the interface between ITO and Cu 2 O, the macroscopic current-voltage characteristics of the structure ITO/Cu 2 O/Ni/Au were studied, revealing no current rectification (not shown).
As observed in Fig. 1, the density of grain boundaries is significantly increased towards the substrate interface. Due to the pinned Fermi level in CuO and the degeneracy of ITO, the ITO/CuO junction does not show current rectification. 19 This means that the current which is probed at the surface of a Cu 2 O grain is shunted by the CuO in the grain boundaries in the vicinity to the substrate interface. This is in accordance with the high conductivity observed in copper oxide films at low thicknesses. 18 As a consequence, the rectification observed in Fig. 2(e) is related to the current transport over the interface between the copper oxide surface and the AFM tip. This is in line with the absence of a CuO structure on the surface of the samples; see Fig. 1(a). Since the Cu(ii) oxidation state is detected by XPS of air-exposed Cu 2 O samples, 11,14,16 the copper oxidation must be mainly caused by the adsorbates from the ambient air.
The work function of the platinum-iridium coating of the AFM tip is 5.3-5.5 eV, Cu 2 O has a work function of around 5.0 eV, and CuO of has a work function of 4.7-5.5 eV. 19,24,25 Hence, based on the Schottky-Mott rule, none of the copper oxides should form a barrier with the AFM tip material. Most probably, surface states of Cu 2 O cause the observed barrier, but the exact origin cannot be clarified here.
The Fermi level in the CuO grain boundary is pinned, which allows no band bending at the unbiased contact between AFM tip and CuO. The upper limit for the hole barrier height is given by the Fermi level with respect to the valence band maximum in CuO of 0.1 eV. 18,19 Contrarily, Cu 2 O is able to show considerable band bending and can form Schottky barrier heights of 0.5 eV with a wide depletion region with respect to the film thickness. 11,26 Figures 2(f) and 2(g) schematically illustrate the local energy band alignment between the AFM tip and the different copper oxides under reverse bias (negative voltage). A rough estimation of the dominant current transport mechanism can be made by means of comparing the tunneling parameter E 00 [see Eq. (1)] with the thermal energy kT, where k is the Boltzmann constant and T is the temperature, 24,27 Here, q is the charge, is the reduced Planck constant, N is the intrinsic acceptor concentration, m * is the effective mass, and is the dielectric constant. The condition for tunneling E 00 kT implies a minimum acceptor concentration at a given temperature, here room temperature. For the case of tunneling in CuO with m * = 7.9 m 0 and = 10.5 0 , 20,28 N 2 × 10 20 cm −3 . m 0 is the free electron mass at rest and 0 is the vacuum permittivity. Acceptor concentrations in CuO are not available in the literature. The reported hole carrier concentrations at room temperature range between 10 19 cm −3 and 6 × 10 20 cm −3 . 19,20 Based on these values, it can be assumed that the condition for tunneling is fulfilled since the acceptor concentration is always equal or higher than the hole concentration. Hence, tunneling can explain the exponential voltage dependence in reverse bias, which is observed in Fig. 2(e). Due to the high intrinsic acceptor concentration of the material, the barrier at the interface to the AFM tip is very thin, which is schematically illustrated in Fig. 2(f).
For Cu 2 O, the condition for tunneling is N 7 × 10 18 cm −3 using m * = 0.58 m 0 and = 7 0 . 29,30 Reported acceptor concentrations range from 10 14 cm −3 to 10 18 cm −3 . 4,26,31,32 Hence, thermionic emission is the dominant mechanism for current transport in the reverse direction. Due to the high barrier in comparison to the thermal energy at room temperature and the wide depletion region in Cu 2 O, the current in the reverse direction remains below the detection limits of the instrument; see Fig. 2(e). The efficient current blocking is schematically illustrated in Fig. 2(g).
In conclusion, the existence of nanometric CuO between the grains of Cu 2 O was visualized by TEM high resolution automated phase and orientation mapping. The electric properties of the thin films were deconvoluted into the contributions of grains and grain boundaries by conductive AFM. A high grain boundary density causes junction characteristics to be dominated by shunting paths of nanocrystalline CuO and only its absence allows us to observe the rectification expected from Cu 2 O. The presented results obtained from Cu 2 O films deposited by AA-MOCVD corroborate previously published data on Cu 2 O prepared by reactive radio-frequency magnetron sputtering. 18 As a consequence, it can be assumed that the existence of highly conductive CuO-containing grain boundaries in Cu 2 O is a general characteristic of the material, independent of the deposition technique. Hence, to unlock the potential performance of this p-type oxide semiconductor, strategies to control/inhibit the formation of CuO in its grain boundaries must be considered.