Epitaxial Strain-Induced Growth of CuO at Cu2O/ZnO InterfacesA. E. Gunnæs,*,† S. Gorantla,† O. M. Løvvik,†,‡ J. Gan,† P. A. Carvalho,‡ B. G. Svensson,†

E. V. Monakhov,† K. Bergum,† I. T. Jensen,‡ and S. Diplas‡

†Department of Physics/Center for Materials Science and Nanotechnology, University of Oslo, P.O. Box 1048, Blindern, N-0316Oslo, Norway‡SINTEF Materials and Chemistry, P.O. Box 124, Blindern, Forskningsveien 1, N-0314 Oslo, Norway

ABSTRACT: Cu2O/ZnO has been envisaged as a potentialmaterial system for the next-generation thin film solar cells.Thus far, the experimental efforts to obtain conversionefficiencies close to the theoretically predicted value havefailed. Combining aberration-corrected (scanning) trans-mission electron microscopy and density functional theorymodeling, we studied the interfaces between single-crystal c-axis-oriented ZnO and high-quality magnetron-sputteredCu2O films. Strikingly, our study shows that the first ∼5 nmof the Cu−oxide films has the structure of the monoclinic CuO phase. The CuO layer is textured with the (111), (111 ), (1 11),(11 1) (1 11), (11 1), (1 11,) and (100) planes parallel to the (0001) and (0001 ) ZnO interfaces. The ionic arrangement on theseplanes resembles the hexagonal arrangement of the ZnO interface, and epitaxy exists across the interface. A continued epitaxialgrowth of [111]-oriented Cu2O follows resulting in epitaxial 180° rotation twins in the Cu2O layer. For the case with the(100)CuO interfacial plane we have (111)[11 0]Cu2O∥(100)[011]CuO∥(0001)[112 0]ZnO. Because of a closer lattice matching ofCuO with ZnO and Cu2O, the total strain and energy is reduced compared to a pure (111)Cu2O∥(0001)ZnO interface. Theexistence of CuO is anticipated to be a contributing factor for the low conversion efficiencies obtained experimentally.

■ INTRODUCTION

Cuprous oxide or cuprite (Cu2O) is a low-cost and nontoxic“intrinsic” p-type semiconductor with a band gap of ∼2.1 eV.Combined with n-type ZnO, which has a band gap of 3.4 eV, aCu2O/ZnO p−n heterojunction would be obtained. Cu2O hasa primitive cubic crystal structure (Pn3m; a = 0.427 nm) wherethe Cu+ and O2− ions are located on fcc and bcc sublattices,respectively. ZnO has a unit cell with a hexagonal wurzitecrystal structure (P63mc; a = 0.325 nm, c = 0.521 nm). Aheteroepitaxial relationship between ZnO and Cu2O has beenexpected due to a relatively small lattice mismatch of 7.6%anticipated for the Cu2O(111)/ZnO(0001) interface.

1

A most interesting prospect for the Cu2O/ZnO materialssystem is related to its theoretically predicted conversionefficiency of 18%.2 With its low cost, nontoxic constituents, andpotentially high conversion efficiency it has been envisaged as apotential materials system for the next-generation thin-film-based solar cells.3−6 At the present, however, the highestreported conversion efficiencies are only 2% for Cu2O/ZnOheterojunction thin films3 and 4.12% and 5.38% when anundoped Ga2O3 or ZnO thin interfacial layer is present inAZO/n-semiconductor/p-Cu2O heterojunctions.7,8 Hence,there is an increasing effort to find solutions to improve theconversion efficiency.The majority of the research efforts in this field are, however,

focused on developing synthesis methods, and only fewinvestigations are devoted to in-depth atomic-scale inves-tigations.1,9−11 In order to obtain a fundamental understandingabout the reasons behind the discrepancy between the

theoretical and the experimentally obtained conversionefficiencies of Cu2O/ZnO heterojunction solar cells, modelsystems of high-quality Cu2O films on O- and Zn-polar single-crystal ZnO were used in the present study. A combination ofatomically resolved scanning transmission electron microscopy(STEM), electron diffraction (ED), and density functionaltheory (DFT) based modeling was employed to investigate thematerial system and, in particular, the substrate−film interface,as the performance of thin film-based devices is well known tobe largely affected by the quality of its interfaces.

■ EXPERIMENTAL SECTION

The Cu2O films investigated in the present study were grownby direct current/radio frequency (dc/rf) reactive magnetronsputtering on c-axis-oriented single-crystal ZnO waferspurchased from Tokyo Denpa Co., Ltd. Detailed descriptionof the deposition parameters is reported by Gan et al.12 In brief,the substrate wafers were double-side polished and pretreatedto make sure they were free of any contamination before thesputtering deposition. One side of such a prepared wafer servedas the Zn-polar substrate while the other side as the O-polarsubstrate. During the magnetron sputtering deposition, thesubstrate temperature was maintained at 400 °C and the basepressure of the chamber was below 4 × 10−4 Pa.

Received: July 18, 2016Revised: September 13, 2016Published: September 19, 2016

Article

pubs.acs.org/JPCC

© 2016 American Chemical Society 23552 DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

Cross-section TEM specimens were prepared by theconventional tripod wedge mechanical polishing protocol.The final thinning of the specimens was done by ion millingfor 1 h in a Fischione 1010 instrument with beam energy andcurrent of 5 kV and 5 mA, respectively, and milling angle of±6° on both sides of the specimen.A JEOL 2100F microscope operated at 200 kV and an FEI

Titan G2 60-300 microscope equipped with a DCOR probeCs-aberration corrector, operated at 300 kV, were used in thepresent study. STEM high-angle annular dark field (HAADF)Z-contrast imaging was performed using the FEI microscopewith a probe current of ∼100 pA and probe convergence andcollection angles of 22 and 76−200 mrad range, respectively.The resulting spatial resolution was ∼0.08 nm. HAADF imagesshow primarily atomic number (Z) contrast, and in the presentcase, high-resolution (HR) HAADF images offer advantageover HRTEM ones owing to the direct correlation between theimage contrast and the cation positions in the sample. FastFourier transform (FFT) analysis was used on HR images, andlattice strain measurements were carried out using thegeometric phase analysis (GPA) plugin for the DigitalMicro-graph software.13

The DFT calculations were performed with the plane-wave-based VASP code14,15 using the Perdew−Burke−Ernzerhofgradient approximation16 and the projector augmented wavemethod.17 The plane-wave cutoff was 400 eV, and the k-pointdensity was at least 4 points per Å−1 in each unit cell direction.The force relaxation criterion was 0.05 eV/Å, and the twooutermost layers of one side of the slab structure were fixed.The interface structures were built using the ASE environ-ment.18 The strain was divided equally between the two sides ofthe interface, and a 5 × 5 two-dimensional potential energysurface (PES) was generated by lateral translations of one of thesides relative to the other. Each PES point was obtained byrelaxation of the whole interface structure, and the lowestenergy was taken to represent the interface. A 1 nm thick layerof vacuum was inserted between the two phases on one side, soonly one interface was formed. For reference energies, anadditional layer of 1 nm vacuum was inserted between the twophases on the other side, creating two separate slabs divided byvacuum on both sides. The interface energy was defined as theenergy difference between the interface model and thereference model.

■ RESULTS AND DISCUSSIONFigure 1 shows overview images of the cross sections of the (a)O-polar Cu2O/ZnO and (b) Zn-polar Cu2O/ZnO specimens.The Cu2O films are in both cases dense and of high quality.Prior XRD analysis of the specimens showed strong (111)Cu2Oand (0002)ZnO diffraction peaks12 consistent with previousreports where it has been interpreted as evidence for aheteroepitaxial relationship between Cu2O and ZnO due to the7.6% mismatch between the two lattices.1

Representative selected area diffraction (SAD) patterns fromthe Cu2O films and the ZnO substrates are shown in Figure 2.The SAD pattern from the Cu2O films in Figure 2c can beindexed in agreement with a single-crystal zone axis pattern([11 2]). However, the complex pattern seen in Figure 2f canbe described as two single-crystal zone patterns ([1 10] and[110]) twin related with a 180° rotation around the common[111] axis. Such a twin relation has previously been reportedfor Cu2O nanoparticles and in Cu2O film upon oxidation of asingle-crystal Cu.19−21 The additional reflexions located 1/3

and 2/3 between {111} rows are due to multiple scattering, i.e.,the two twin regions are partly overlapping and the electronswill, due to their short scattering mean free path, scatter fromboth twin orientations resulting in additional reflexions. Fromthe indexed patterns in Figure 2, it can be concluded that thesubstrate and the Cu2O films has the two orientation variants:( 1 1 1 ) [ 1 1 0 ] C u 2 O ∥ ( 0 0 0 1 ) [ 1 1 2 0 ] Z n O a n d ( 1 1 1 )[1 10]Cu2O∥(0001)[112 0]ZnO.The orientation relationship (111)[11 0]Cu2O||(0001)

[112 0] ZnO is in agreement with a heteroepitaxial relationshipbetween the two phases reported by others and attributed toenergetically favorable periodic lattice coincidence betweenthese two crystal lattices.10 Wang et al.22 reported that theatomic arrangement of the first atomic layer of Cu2O predictsthe atomic arrangement throughout the film. This would implythat one get growth of textured Cu2O films but not necessarilyan epitaxial relationship between Cu2O and ZnO. The observedtwinned growth of Cu2O is, however, regarded as evidence ofepitaxial growth. As shown in Figure 3, the Cu ions across thegrain boundary have a changed stacking order of their {111}Cu planes (B, C, A, B, C... and C, B, A, B, C...). This isconsistent with two different starting growth sites on a substraterepresenting an “A” layer. Such twins can be described asepitaxial twins and have been reported for other oxide-basedepitaxial heterostructures.23

Figure 1. (a) HAADF STEM cross-section overview image of the O-polar Cu2O/ZnO, and (b) DF TEM image of the Zn-polar Cu2O/ZnO films showing contrast variations in the Cu2O films consistentwith polycrystalline films.

Figure 2. SAD patterns from two specimen orientations where theZnO substrate is oriented along the (a−c) [11 00] and (d−f)[112 0]ZnO ZA. (a−c) (111)[11 0]Cu2O||(0001)[112 0]ZnO orientation

relationship; (e and f) Cu2O twin relationship.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

23553

The XRD of the current specimens showed that the cellparameters of the Cu2O films were 0.4318 and 0.4301 nm forthe O- and Zn-polar substrates, respectively.12 This shows thatthe lattice is only 1.1% and 0.7% strained relative to unstrainedCu2O (a = 0.427 nm). The twin growth could be assumed tohave relaxed the strain induced by epitaxy. However, structuraldefects located close to the heterointerface could also cause thestrain in the grown film to be reduced.Figure 4 shows representative HAADF STEM images from

Cu2O/ZnO interfacial regions where bright spots correspondto Zn- and Cu-atom positions in the substrate and film region,respectively. Independent of substrate polarity, the first ∼5 nmof the grown Cu−O films show distinct different atomicarrangement and grain sizes (5−10 nm in diameter) compared

to the much larger twin domains of the Cu2O film above(Figure 4a and 4b).The XRD analysis and SAD from interfacial regions (Figure

2) did not show indications of secondary phases. In both casesthis can be explained by the relative low volume fraction of thesecondary phase and polycrystalline structure and in the case ofXRD small grain sizes. However, X-ray energy-dispersivespectroscopy (EDS) and fast Fourier transform (FFT) imageanalysis of HR STEM images, as illustrated in Figure 4,performed at various positions along the interfacial regionsshowed that the interfacial layer was not consistent with Cu2O.However, the monoclinic CuO phase with space group C1c1and lattice parameters a = 0.469 nm, b = 0.343 nm, c = 0.51370nm, and β = 99.546°24 did give a good fit to the FFT data andis consistent with the EDS data. This is unexpected as the CuOlayer was grown under oxygen-poor deposition conditionsoptimized for Cu2O growth.On the basis of the FFT analysis of the HR STEM data, it is

clear that the CuO grains grow on the ZnO substrate withpreferred orientation relationships. Figure 4 illustrates the(111)[11 0]Cu2O∥(100)[011]CuO∥(0001)[1120]ZnO orientationrelationship which together with (0001)ZnO∥(111), (111 ),(1 11), (11 1 ) (1 11), (11 1), (111 )CuO∥(111)Cu2O represents thecommon interfaces found. All CuO planes ((111), (1 11),(11 1), (11 1) (11 1), (1 11), and (100)) represent dense planesof Cu ions with Cu arrangements similar, but distorted, to theone present in {111}Cu2O planes. In the {111}Cu2O planes theCu−Cu cation interionic bond length (CIBL) is 0.301 nm. TheCu−Cu CIBL values at the CuO interfaces are listed in Table 1,

Figure 3. (a) STEM image illustrating the stacking order of A,C,B,A...and A,B,C,A... in the twinned Cu2O grains. (b) Illustration of thepossible starting points with respect to orientation of the Cu2O latticegiving rise to 180° rotation twins where the green spheres representCu positions.

Figure 4. STEM-HAADF images from the (a and b) O-polar and (c) Zn-polar Cu2O/ZnO interfaces where the corresponding FFT patterns from bshow the (111)[11 0]Cu2O∥(100)[011]CuO∥(0001)[112 0]ZnO orientation relationship and d, e, and f the Cu, Zn, and O EDS elemental distributionsin c.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

23554

and the similarity in atomic configuration with {111} planes ofthe face-centered cubic Cu sub lattice of Cu2O (Figure 3) isillustrated for the (100)CuO plane in Figure 5. As such, either(100)CuO or any of the 111 type of planes of CuO ((111),(111 ), (111), (11 1 ) (11 1), (11 1), (1 11)) can act as suitablesurfaces for epitaxial growth of {111}Cu2O, and this behavior canjustify the presence of Cu2O twin-related grains instead ofrandomly oriented (111) textured Cu2O grains one couldexpect with no epitaxial relations.To further rationalize the formation of a CuO interfacial

layer under oxygen-poor deposition conditions, DFT modelingwas performed. A series of stoichiometric CuO/ZnO, Cu2O/ZnO, and CuO/Cu2O interface models were constructed afteran automatic interface model search, based on the ASEenvironment,18 and relaxed to their low-energy configuration.Many hypothetical interface models with relatively small latticemismatch were identified. However, in order to obtain a stableepitaxial interface, atomic columns have to match as much aspossible at the interface plane and the local stoichiometry mustbe appropriate. This means that the most stable interface is notnecessarily the one with the lowest lattice mismatch and that ahypothetical interface with low lattice mismatch may beimpossible to realize in practice. This is demonstrated inFigure 6a and 6b, showing the Cu2O(111)/ZnO(0001)interface with the lowest volumetric strain (3.1%) found withour automatic routine. The low structural compatibility due tononmatching atomic columns indicates that this model is notlikely to form. The model shown in Figure 6c and 6d representsthe potential compatibility between the two phases; however, inthis case the interface exhibits significantly higher strain (7.8%).The interface models that are shown in Figure 6c and 6d

(111) [11 0]Cu2O∥(0001)[112 0]ZnO, Figure 6e and 6f (111)[11 0]Cu2O∥(100)[011]CuO, and Figure 6g and 6h (100)[011]CuO∥(0001)[112 0]ZnO show excellent epitaxy in all

directions and the lowest strain. Supporting the validity ofthe selection scheme, they also represent two of theexperimentally observed interfaces (Figure 4).A comparative evaluation of the stability of the competing

interfaces has been performed based on the calculated interfaceenergy Eint. Eint, defined as the energy required to pull apart thetwo constituting parts leaving two free surfaces, is largest for the

Table 1. Cu−Cu Cation Interionic Bond Lengths (CIBL) on the Interfacial Planes of CuO Based on the CuO LatticeParameters

Cu−Cu (nm) 0.376 0.343 0.318 0.310 0.308 0.291

(100) X X X(111), (111 ), (1 11), (11 1), (1 11) X X X X(11 1), (1 11) X X X X

aa = 0.46927 nm, b = 0.34283 nm, c = 0.51370 nm, and β = 99.546°.24 Corresponding CIBL for ZnO and Cu2O are 0.325 and 0.301 nm,respectively.

Figure 5. Projection of (100) planes in CuO, where blue and redspheres represent Cu and O ions, respectively. Crossed and theuncrossed Cu ions are on alternating planes. Arrows representinterionic Cu−Cu distances: 0.308 nm (black), 0.310 nm (orange),and 0.343 nm (red) based on ref 24.

Figure 6. Energetically relaxed interface model structures from DFTcalculations showing (a−d) (111)Cu2O||(0001)ZnO, (e and f)(111)Cu2O||(100)CuO, and (g and h) (100)CuO||(0001)ZnO interfaces,seen in the (a, c, g) [1100]ZnO, (b, d, h) [1120]ZnO, (e) [1 12]Cu2O, and(f) [1 10]Cu2O projections. (a and b) Interface with the lowest strainobtained by automatic routine, showing very low structuralcompatibility.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

23555

most stable interface. The resulting DFT-calculated interfaceenergy Eint showed that both the (100)CuO/(0001)ZnO and the(111)Cu2O/(100)CuO display relatively high interface energies(2.5 and 3.1 eV), whereas the (111)Cu2O/(0001)ZnO interfaceenergy is lower (1.5 eV). In other words, the hypothetical(111)Cu2O/(0001)ZnO interface is thermodynamically relativelyunstable, favoring the presence of a CuO interfacial layer.As seen from Table 2, the main difference between the

interface models is related to the strain between the

constituting phases. The strain is very high in the case of(111)Cu2O/(0001)ZnO; the linear strain is 7.8%, and the volumestrain is 16.2%. This is associated with a large energy penalty,and it also means that if such an interface is formed, it wouldmost likely abound with defects (such as dislocations) whichmay tend to relieve the build up of the stresses.With a calculated volumetric strain of 5.8%, the (100)CuO/

(0001)ZnO interface is found to be much more favorable andhence gives a strong impetus to form epitaxial CuO instead ofCu2O on top of ZnO. Nevertheless, the low oxygen contentduring synthesis means that there is a thermodynamic drivetoward the formation of Cu2O. Thus, even if the epitaxialgrowth of CuO is preferred over Cu2O at the ZnO surface, it isreasonable to assume that Cu2O will start forming when strainis reduced to a given level during CuO growth. CuO is in thisway acting as a buffer layer accommodating the strain betweenZnO and Cu2O.

The picture that emerges is the following: instead of a directinterface between ZnO and Cu2O as depicted in Figure 6a and6b, CuO forms an intermediate layer to accommodate thestrain between ZnO and Cu2O. The latter situation can beillustrated by combining Figure 6c and 6e or alternativelyFigure 6d and 6f. This corresponds very closely to theexperimental micrograph in Figure 4b.The strain at the (100)CuO/(0001)ZnO interface was further

analyzed experimentally by geometric phase analysis (GPA).The GPA εxx map in Figure 7b shows the local strain at theCuO/ZnO interface. The color scheme clearly shows that theCuO lattice is mostly unstrained and that the CuO/ZnO latticemismatch is relaxed by multiple misfit dislocations distributedalong the interface and in its vicinity. Indeed, the measuredCuO lattice is 4.8 ± 1.7% larger than the ZnO lattice along[1 100], which is in agreement with the expected unrelaxedtheoretical 5.8% deviation for unstrained conditions. Theseresults demonstrate that strain is minor in the CuO layer andmostly accumulated in the ZnO layer as there is much strongervariation of εxx contrast below the interface (Figure 7e).An analysis of the Burgers vectors in Figure 7a was

performed using the right-hand start−finish convention. Alldislocations were found to be 90° edge dislocations withBurgers vector 1/2 [1100]ZnO and parallel to the interface.25,26

The white and red arrows in Figure 7 correspond to the extrahalf plane in the ZnO lattice at (1101 ) and (1 101) planes asseen in the Bragg-masked inverse FFT images (Figure 7f and7g) where the two dislocation components of the 90° full edgedislocation can be clearly seen.The dislocation cores are spaced ∼4 nm apart along the

interface (Zn dislocation core spaced with 14th, 16th, and 19thZn−Zn CIBL). The (100)CuO plane has in its unstrained state,as illustrated in Figure 5 and listed in Table 1, Cu−Cu CIBLvalues of 0.308 (black), 0.310 (orange), and 0.343 nm (red). Ina [001]CuO projection, the Cu−Cu CIBL alternates between0.308 and 0.310 nm along the unstrained CuO/ZnO interfacewith an average Cu−Cu CIBL of 0.309 nm. An unstrained ZnO

Table 2. Strain and Calculated Interface Energy from theDFT Study where Vectors 1 and 2 Represent in-PlaneAtomic Bond Directions at the Interfaces

strain (%)

interface vector 1 vector 2 volume Eint (eV)

(100)CuO/(0001)ZnO 0.47 5.34 5.84 2.50(111)Cu2O/(0001)ZnO 7.80 7.80 16.22 1.47(111)Cu2O/(100)CuO 5.01 1.64 6.74 3.14

Figure 7. (a and d) HRSTEM image showing burgers circuits around misfit dislocations at the CuO/ZnO interface, (b and e) corresponding GPAεxx map, (c) FFT of a indicating the spots used for GPA, and (f and g) inverse FFT moire images enhancing the (1 101 ) and (1 101) ZnO planesindicated by white and red arrows in d.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

23556

lattice has, correspondingly, a CIBL of 0.325 nm in the (001)plane. Hence, in an unstrained state, along a (100)CuO/(0001)ZnO interface, one has that ∼19* Zn−Zn CIBL = 18*averaged Cu−Cu CIBL. This is in good agreement with theobservation of the dislocation spacing in Figure 7 pointing tototal low strain in the CuO lattice.The presence of the misfit dislocations along the CuO/ZnO

interface reduces the strain relative to the DFT-calculated value,where no dislocations were taken into account and increasingthe driving force to grow CuO even more. It is thus clear thatattempts to form an epitaxial layer of Cu2O directly on top ofZnO (0001) are likely to fail. This can be understood from thecubic crystal structure of Cu2O, which gives very few possiblecombinations of cell parameters (and atomic column arrange-ments). CuO has, on the other hand, a monoclinic crystalstructure with three different lattice constants and can thusprovide a large variety of interface unit cell sizes and shapes.This flexibility turns out to be more important than processingparameters when it comes to growing an epitaxial Cu−O filmon top of ZnO. Thus, a path toward epitaxial Cu2O on ZnOcould be through an electronically benign buffer layersuppressing the formation CuO, as corroborated by animproved conversion efficiency of Cu2O/ZnO heterojunctionsolar cell having an intermediate interfacial layer.7,8

■ CONCLUSIONSBy atomically resolved STEM and FFT analysis it was revealedthat single-crystalline (0001) and (0001 ) ZnO substrates havean unintentional ∼5 nm thin intermediate layer of CuO at theCu2O/ZnO interfaces irrespective of ZnO surface polarity. Thegrowth of CuO is textured on the ZnO substrates, showingevidence of a continuation of densely packed CuO planes ofcations such as (111), (111 ), (111), (11 1 ), (11 1), (11 1), (1 11),and (100) parallel with the (0001)ZnO, (0001 )ZnO, and(111)Cu2O interfaces. The densely packed CuO planes arefound to act as a template, facilitating growth of epitaxialtwinned Cu2O films.The DFT study of the theoretical (111)[11 0]Cu2O∥(0001)-

[1120]ZnO and the experimentally observed (100)[011]CuO∥(0001)[112 0]ZnO and (111)[11 0]Cu2O∥(100)[011]CuO inter-faces showed that both the strain and the interface energy werein favor of epitaxial growth of CuO on the ZnO (0001) surface.GPA showed evidence of localized strain associated with 90°

edge misfit dislocations with the Burgers vector 1/2 [1100]ZnOalong the (100)CuO/(0001)ZnO interface. The misfit dislocationsreduce the strain in the CuO layer, making it even moreenergetically favorable to grow CuO relative to Cu2O.The CuO interlayer, not resolved by XRD, is anticipated to

be inherent to the interface between crystalline Cu2O and ZnOfilms in heterojunction structures. The CuO layer is highlydetrimental for the current rectification and open-circuit voltageof the Cu2O/ZnO heterojunction due to the differences in theband alignment and the band gap (2.1 vs 1.2 eV) and can beone of the causes for the poor conversion efficiency obtainedexperimentally for Cu2O/ZnO solar cells relative to thatpredicted theoretically.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: eleonora@fys.uio.no. Phone: +47 22852812.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

This work was conducted under the research project ES483391Development of a Hetero-Junction Oxide-Based Solar CellDevice (HeteroSolar), financially supported by the ResearchCouncil of Norway (RCN) through the RENERGI program.

■ REFERENCES(1) Fariza, B. M.; Sasano, J.; Shinagawa, T.; Nakano, H.; Watase, S.;Izaki, M. Electrochemical Growth of (0001)-n-ZnO Film on (111)-p-Cu2O Film and the Characterization of the Heterojunction Diode. J.Electrochem. Soc. 2011, 158, D621−D625.(2) Loferski, J. J. Theoretical Considerations Governing the Choiceof the Optimum Semiconductor for Photovoltaic Solar EnergyConversion. J. Appl. Phys. 1956, 27, 777−784.(3) Mittiga, A.; Salza, E.; Sarto, F.; Tucci, M.; Vasanthi, R.Heterojunction Solar Cell with 2% Efficiency Based on a Cu2OSubstrate. Appl. Phys. Lett. 2006, 88, 163502−1−163502−2.(4) Fay, S.; Kroll, U.; Bucher, C.; Vallat-Sauvain, E.; Shah, A. LowPressure Chemical Vapour Deposition of ZnO Layers for Thin-FilmSolar Cells: Temperature-Induced Morphological Changes. Sol. EnergyMater. Sol. Cells 2005, 86, 385−397.(5) Rai, B. P. Cu2O Solar Cells: A Review. Sol. Cells 1988, 25, 265−272.(6) Takahashi, A. Next Generation Materials Created by Nano-technologies. Erekutoronikusu Jisso Gakkaishi 2003, 6, 1−3.(7) Minami, T.; Nishi, Y.; Miyata, T.; Nomoto, J. High-EfficiencyOxide Solar Cells with ZnO/Cu2O Heterojunction Fabricated onThermally Oxidized Cu2O Sheets. Appl. Phys. Express 2011, 4, 062301.(8) Minami, T.; Nishi, Y.; Miyata, T. High-Efficiency Cu2O-BasedHeterojunction Solar Cells Fabricated Using a Ga2O3 Thin Film as N-Type Layer. Appl. Phys. Express 2013, 6, 044101.(9) Izaki, M.; Shinagawa, T.; Mizuno, K.; Ida, Y.; Inaba, M.; Tasaka,A. Electrochemically Constructed p-Cu2O/n-ZnO HeterojunctionDiode for Photovoltaic Device. J. Phys. D: Appl. Phys. 2007, 40,3326−3329.(10) Jeong, S.; Aydil, E. S. Heteroepitaxial Growth of Cu2O ThinFilm on ZnO by Metal Organic Chemical Vapor Deposition. J. Cryst.Growth 2009, 311, 4188−4192.(11) Wong, L. M.; Chiam, S. Y.; Huang, J. Q.; Wang, S. J.; Pan, J. S.;Chim, W. K. Growth of Cu2O on Ga-doped ZnO and their InterfaceEnergy Alignment for Thin Film Solar Cells. J. Appl. Phys. 2010, 108,033702.(12) Gan, J.; Gorantla, S.; Riise, H. N.; Fjellvag, Ø. S.; Diplas, S.;Løvvik, O. M.; Svensson, B. G.; Monakhov, E. V.; Gunnæs, A. E.Structural Properties of Cu2O Epitaxial Films Grown on c-axis SingleCrystal ZnO by Magnetron Sputtering. Appl. Phys. Lett. 2016, 108,152110.(13) Roesner, H.; Koch, C. T.; Wilde, G. Strain Mapping Along Al-Pb Interfaces. Acta Mater. 2010, 58, 162−172.(14) Kresse, G.; Hafner, J. Ab Initio Molecular Dynamics for LiquidMetals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561.(15) Kresse, G.; Furthmuller, J. Efficient Iterative Schemes for AbInitio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys.Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186.(16) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized GradientApproximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868.(17) Blochl, P. E. Projector Augmented-Wave Method. Phys. Rev. B:Condens. Matter Mater. Phys. 1994, 50, 17953−17979.(18) Bahn, S. R.; Jacobsen, K. W. An Object-Oriented ScriptingInterface to a Legacy Electronic Structure Code. Comput. Sci. Eng.2002, 4, 56−66.(19) Yin, M.; Wu, C.; Lou, Y.; Burda, C.; Koberstein, J. T.; Zhu, Y.;O'Brian, S. Copper Oxide Nanocrystals. J. Am. Chem. Soc. 2005, 127,9506−9511.(20) Li, L.; Wang, J.; Wang, R.; Liu, H.; Jia, C.; Ma, L.; Yu, Y.Atomistic Study on Twinning of Cu2O Quantum Dots. Appl. Phys.Lett. 2006, 89, 113109.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

23557

(21) Li, J.-Q.; Mei, Z.-X.; Ye, D.-Q.; Hou, Y.-N.; Liu, Y.-P.;Kuznetsov, A. Yu.; Du, X.-L. Temperature Dependence of Cu2OOrientations in Oxidation of Cu (111)/ZnO (0001) by OxygenPlasma. Chin. Phys. B 2012, 21, 076401.(22) Wang, Y.; Ghanbaja, J.; Soldera, F.; Boulet, P.; Horwat, D.;Mucklich, F.; Pierson, J. F. Controlling the Preferred Orientation inSputter-Deposited Cu2O Thin Films: Influence of the Initial GrowthStage and Homoepitaxial Growth Mechanism. Acta Mater. 2014, 76,207−212.(23) Qiao, L.; Xiao, H. Y.; Weber, W. J.; Biegalski, M. D. Coexistenceof Epitaxial Lattice Rotation and Twinning Tilt Induced by SurfaceSymmetry Mismatch. Appl. Phys. Lett. 2014, 104, 221602.(24) Åsbrink, S.; Waskowska, A. CuO: X-ray Single-Crystal StructureDetermination at 196 K and Room Temperature. J. Phys.: Condens.Matter 1991, 3, 8173−8180.(25) Kim, T. W.; Lee, D. U.; Lee, H. S.; Lee, J. Y.; Park, H. L. StrainEffects and Atomic Arrangements of 60° and 90° Dislocations Nearthe ZnTe/GaAs Heterointerface. Appl. Phys. Lett. 2001, 78, 1409−1411.(26) Li, J.; Zhao, C.; Xing, Y.; Su, S.; Cheng, B. Full-Field StrainMapping at a Ge/Si Heterostructure Interface. Materials 2013, 6,2130−2142.

The Journal of Physical Chemistry C Article

DOI: 10.1021/acs.jpcc.6b07197J. Phys. Chem. C 2016, 120, 23552−23558

23558