Available online at www.sciencedirect.com

ScienceDirectActa Materialia 85 (2015) 122–131

www.elsevier.com/locate/actamat

Role of materials chemistry on the electrical/electronicproperties of CuO thin films

Yonglong Shen,a Meilan Guo,a,b Xiaohong Xiaa,b and Guosheng Shaoa,c,⇑aInstitute for Renewable Energy and Environmental Technologies, University of Bolton, Bolton BL35AB, UK

bSchool of Materials Science and Engineering, Hubei University, Xueyuan Road, Wuhan 430062, People’s Republic of ChinacUK-China Centre for Multi-functional Nanomaterials, Zhengzhou University, Zhengzhou 450001, People’s Republic of China

Received 6 August 2014; revised 2 November 2014; accepted 9 November 2014

Abstract—CuO thin films with different levels of compositional deviation from 50:50 stoichiometry have been fabricated using radiofrequency sput-tering deposition wherein the sputtering gases consisted of oxygen and argon in various proportions. The microstructures of the thin films were char-acterized by combining a series of advanced methods including X-ray diffraction, energy dispersive X-ray analysis, scanning and high-resolutiontransmission electron microscopy, electron energy loss spectroscopy, and high-resolution X-ray photoelectron spectroscopy. The results showed thatthe chemical states of Cu and O in the thin films depended upon CuO composition and sputtering conditions, so that different levels of Cu vacanciesdictated the electrical/electronic properties of the thin films. The ability to control the compound composition and associated alloying chemistryenables tuning of the concentration and mobility of holes in CuO, hence creating alow-cost and environmentally friendly semiconductor from abun-dant materials. This offers an essential technical basis in engineering photonic devices such as pn or Schottky diodes, thus opening new avenues foreconomic harvest of solar energy using diodes solely based on sustainable oxides.� 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Copper oxides; Alloy chemistry; Hall measurement; Current–voltage characteristics; Schottky junction

1. Introduction

Interest in the Cu–O system started from the early yearsof semiconductor physics, with many experimental phe-nomena and applications having been originally discoveredor demonstrated in Cu2O, the cuprous oxide. This cubiccompound exhibits mostly p-type semiconducting charac-teristics, with a band gap of 2.1–2.6 eV [1–3]. There hasbeen continued interest in studying this cuprous oxidematerial for various applications such as photonic devices[2], superconductivity [4,5] and photovoltaic (PV) harvestof solar energy [6–8]. Another copper oxide in the Cu–Osystem is the cupric oxide CuO, a monoclinic p-type semi-conductor with a reported narrow band gap of 1.10–1.71 eV [9–11].

In recent years, CuO has attracted increasing attentiondue to its interesting electrical and optical properties [12–14], together with its low cost and environmental friendli-ness. Such experimental work has demonstrated greatpotential in using CuO for a wide range of applicationssuch as gas sensing [15], catalysis [16], field emission[17,18] and PV cells [6,19]. While previous work

http://dx.doi.org/10.1016/j.actamat.2014.11.0181359-6462/� 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights

⇑Corresponding author at: Institute for Renewable Energy andEnvironmental Technologies, University of Bolton, Bolton BL35AB,UK; e-mail: G.Shao@bolton.ac.uk

[15,16,20–22] on CuO mostly focused on its catalytic andgas-sensing properties, PV cells based on CuO havereceived more attention recently because of the suitableband gap [9,19] of CuO and its easy fabrication into thinfilms or nanowires [23,24]. The band gap of CuO is closeto that of Si [25] and GaAs [26], and in theory the achiev-able solar conversion efficiency could be up to 33% for asingle-junction PV cell with a CuO band gap of �1.4 eV[27–30]. Recent theoretical modeling has shown that PVcells with even higher efficiency can be achieved by utilizingCu2O, CuO and wide-gap oxides in a WAV (window/absorber/voltage-enhancer) architecture [28,29]. It is envis-aged that such oxide PV cells can offer a desirable techno-logical means to reduce the cost of PV cells remarkably,while using non-toxic materials with sustainable resources[28,29,31].

However, experimental realization of CuO-based PVcells has proved very difficult, and to date only 0.41% effi-ciency has been achieved for heterojunction PV cells basedon CuO/Si thin films [19]. We have shown from numericalanalysis that one needs to be able to tune the optical andelectrical properties of copper oxides to maximize theirpotential for PV energy harvest [28–30]. It is essential tobe able to control the charge carrier concentrations in cop-per oxides, so that a suitable potential profile around the p–n junction(s) can be realized to confine the depletion region

reserved.

Y. Shen et al. / Acta Materialia 85 (2015) 122–131 123

in the CuO layer for optimized production of electron–holepairs [28–30]. Fundamentally, these properties of oxidesemiconductors depend strongly on the materials chemistryassociated with native defect concentration, which in turndepends strongly on the fabrication conditions and partic-ularly the achievable oxygen contents [32–36].

A systematic study of the control of hole concentrationand the associated electrical properties is still lacking,despite efforts having been made to synthesize cupric mate-rials using various methods such as reactive sputteringdeposition [37,38], wet chemical synthesis [39], thermalevaporation [9], sol–gel processing [40], etc. It is widely rec-ognized that for large area applications such as thin-filmPV cells, the reactive sputtering deposition technique standsout for process control and overall material and interfacialqualities.

In this work, we attempt for the first time to investigatethe effect of materials chemistry associated with oxygencontent in sputtering deposited CuO thin films, aiming atcontrolling the optoelectrical properties essential for thefabrication of high-efficiency PV cells. The results show thatCuO films with compositions close to the ideal cupric stoi-chiometry have lower hole concentrations than those oflower or higher oxygen contents, and deviation of compo-sitions from the ideal stoichiometry leads to enhanced p-type conductivity through self-doping via increased densityof native defects. In spite of the different oxygen contents,the optical band gaps of CuO thin films in this work areall �1.4 eV, which is within the 1.2–1.6 range for maximumefficiency for solar cells. This ability to control hole concen-tration through tailored oxygen content offers a fundamen-tal materials basis for engineering high-efficiency solar cellsbased on copper oxides.

2. Experimental

CuO thin films were deposited on silicon and glass sub-strates at room temperature by reactive radio frequency(RF) magnetron sputtering using an AJA multi-target sys-tem. Before sputtering deposition, silicon and glass sub-strates were cleaned by a sequence of acetone, isopropylalcohol and deionized water in an ultrasonic bath. The Sisubstrate was B doped p-type wafer with a hole densityof 1014 cm�3 (University wafer Ltd, Part No. S4P01SP,ID 452). The thin surface layer of amorphous SiO2

(�2 nm thick) on the Si wafer was kept in order to avoida seeding effect in the initial stage of coating.

For sputtering, the distance between the Cu target (pur-ity 99.995 wt.%) and substrates was 6.5 cm. The base pres-sure of the deposition chamber was evacuated down to1.07 � 10�3 Pa and the substrates were precleaned byplasma irradiation for 10 min with a 30 W substrate biaspower, in order to ensure the substrate surface was clean.The oxygen level in the reactive sputtering gas was variedby controlling the gas flow rates at different Ar/O2 ratios.A lower range of oxygen percentage (O2/(O2 + Ar)) from13% to 47% was administered by using Ar/O2 ratios of20 sccm/3 sccm, 20 sccm/8 sccm, 20 sccm/10 sccm, 20 sccm/14 sccm and 20 sccm/18 sccm. A higher oxygen percentagerange from 57% to 77% was achieved by keeping theoxygen flow rate at 20 sccm and reducing the argon flowrate from15 to 6 sccm. The deposition duration was either3 h for sputtering at lower oxygen percentage range with

a 120 W RF power, or 20 h for sputtering at the higheroxygen percentage range with a 90 WRF power. Thesubstrates were rotated at 20 rpm during deposition toimprove film uniformity.

X-ray diffraction (XRD) was carried out with a PANa-lytical PERTPRO system using Cu Ka radiation. Scanningelectron microscopy (SEM) was carried out using aJSM6510 system equipped with an Oxford Instruments sys-tem for energy dispersive X-ray analysis (EDX). Forimproved accuracy in quantitative EDX measurement,the Cliff–Lorimer factors were calibrated using standardoxide samples of known compositions (stoichiometricCuO, SiO2). Cross-section samples for transmission elec-tron microscopy (TEM) was prepared using focused ionbeam milling; details are presented elsewhere [41]. TheTEM study was carried out using a Philips CM200 micro-scope with a field-emission source operating at 200 kV.

X-ray photoelectron spectroscopy (XPS) spectra wererecorded by a Thermo VG Multilab 2000 spectrometerunder a vacuum of 3 � 10�7Pa. The radiation source wasAlKa with a power of 300 W. Survey scans were performedat a pass energy of 100 eV and a step size of 1 eV, while apass energy of 25 eV and a step size of 0.05 eV wereadopted for the high-resolution scans of Cu2p3/2, O1s andCu L3VV. The C 1s photoelectron peak of the adventitiouscarbon at 284.6 eV was used as a reference for charge shiftcalibration. The Auger parameter of Cu was calculated byadding the binding energy of photoelectrons for Cu2p3/2

and the kinetic energy of Auger electrons for CuL3VV.The overlapped XPS peaks were deconvoluted, and fittingswere performed using the OriginLab (version 8.0) peak-fit-ting module with the Gaussian peak type. The values of thefull width at half maximum (FWHM) were restricted to beclose for same core-level photoelectron peaks for curvefitting.

Optical absorption properties were measured by a Shi-madzu UV3600 system, covering a spectral range fromultraviolet (UV) to infrared light. Carrier mobility and con-centration were determined by Hall measurement (Sel-TekHMS-3000 system with a magnetic field of 0.5T). The cur-rent–voltage (I–V) characteristics of the samples wererecorded between laterally arranged Cu contacts over thetop surface of oxide films. The Cu contacts were depositedby electron beam evaporation at room temperature, withinsulating glass substrates being used for such oxide filmsto avoid errors due to substrate conductance.

3. Results and discussion

3.1. Microstructural characteristics

Typical XRD patterns for copper oxide films depositedwith different Ar/O2 flow fractions are shown in Fig. 1,wherein the standard XRD powder patterns for bothCuO (PDF No. 80–0076) and Cu2O (PDF No. 77–0199)are shown in the lower panels. The thicknesses of the oxidefilms were in the range of 0.86–1.93 m (measured by cross-sectional SEM). For all samples deposited with the sputter-ing gas containing >45% oxygen, only one peak corre-sponding to the CuO (002) peak is present in the XRDpatterns. Referring to the standard powder CuO pattern,one can see that such dramatic enhancement in one diffrac-tion peak is apparently attributed to the formation of

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

O2 in gas: 13%

(220)(202)

(110)

(-111)

O2 in gas: 47%~77%

2θ

(004)

(002)

CuO

(a)

34.6 35.0 35.5 36.0

(-111)

low O2

high O2

(002)

CuO

(b)

Fig. 1. (a) Typical XRD patterns from CuO thin films deposited withdifferent oxygen percentages in sputtering gas. (b) Enlarged view toshow corresponding peak shift. The bottom panel shows the standardXRD pattern of CuO.

10 20 30 40 50 60 70 8046

47

48

49

50

51

52

Oxy

gen

in fi

lm (a

t.%)

O2 % in gas

Region I Region II Region III

Fig. 2. Oxygen content in CuO thin films deposited with differentoxygen fractions in sputtering gas.

124 Y. Shen et al. / Acta Materialia 85 (2015) 122–131

fibrous texture in the thin films. In the sample depositedwith lower percentages of O2 sputtering gas, the presenceof an additional CuO (110) peak to the left of the majorone can be observed. There are also two additional minorpeaks at 13.0� and 37.6�. None of these peaks can be attrib-uted to CuO, Cu4O3 or Cu2O as they are >1� off the stan-dard peak positions. Within an error of 0.5o, these minorpeaks can be assigned to the Cu64O phase (PDF No. 77–1898, end-centred orthorhombic lattice), as (111) and(026) peaks (13.5� and 37.3�), respectively.

It is interesting to note that the main peak in the low-oxygen film was shifted to a slightly higher angle corre-sponding to the (�111) CuO peak position instead(Fig. 1b). One notes also that the main peak in the low-oxy-gen film is widened considerably, indicating finer grainsizes.

The compositions of oxide films were determined byareal EDX analysis under SEM, using experimentally cali-brated Cliff–Lorimer factors. Fig. 2 shows the resultantoxygen contents vs. the oxygen fraction in the sputteringgas. The error bars in the figure were derived from repeatedmeasurements under the same experimental condition,which is reasonably good for quantifying samples contain-ing light gaseous species such as oxygen. The correlation

between film composition and oxygen level in the sputteringgas can be divided into three regions. In region I, the filmoxygen content was <50% and the resultant oxygen contentin the films was not sensitive to O2 level in the sputteringgas. When the oxygen level in sputtering gas was >30%,the oxygen content in the films began to increase sharplyand film compositions met a plateau of region II, wherethe oxygen contents in the films were about the ideal50:50 stoichiometry of CuO. When the oxygen level in thesputtering gas was >60%, the oxygen content in the filmsbegan to increase further.

From Fig. 2, it can be seen that low oxygen levels in thesputtering gas led to lower oxygen content in the films inregion I, though even in this region the CuO phase was stillthe dominant phase with the strongest peak, albeit themajor peak was shifted to (�111) instead. The presenceof the (110) CuO peak, which is rather weak in the stan-dard powder pattern, in the film with lowest oxygen con-tent, was also attributed to the preferred orientation. Thecoexistence of more than one preferred CuO orientationis believed to be associated with the presence of someCu64O phase in this low-oxygen sample. It is not surprisingthat in samples of higher oxygen content (regions II andIII), only CuO was present in the films, which is consistentwith the Cu–O phase diagram wherein CuO is the phase indirect equilibrium with oxygen, and Cu2O exists betweenCu and CuO when the oxygen concentration in the materialis <50%. The widened composition range in the films con-taining only CuO could be attributed to the far-from-equi-librium processing condition during sputtering deposition,when the substrates were subjected to no additional heatingexcept for heating due to the dissipation of energies in thecondensed ions in the film. In the steady state of the coatingprocesses in this work, the film temperature was estimatedto be <250oC, which was too low to facilitate fullequilibrium.

A change in oxygen content was shown to induce someslight peak shift in the main CuO peak. Fig. 2b shows thatthe position of this main peak in the XRD patterns wasslightly shifted to lower diffraction angles with increasingoxygen content. Consulting the standard powder diffrac-tion pattern of CuO, one can see that the position of themajor peak in samples with higher oxygen contents corre-sponds to the (002) peak (2h: 35.378 vs. PDF card value

Y. Shen et al. / Acta Materialia 85 (2015) 122–131 125

of 35.385). The peak positions in the low-oxygen samplesare closer to that of the (�111) CuO peak (35.55 vs.35.539). As O atoms exist in the interstitial sites in the lat-tice of the metal atoms, one may argue that lowering theamount of oxygen causes some lattice contraction. Thisis, however, in contrast with the constant peak positionsin samples in regions II and III. It is thus more likely thatthe observed shift was due to the different orientation pref-erence in the low- and high-oxygen films, i.e. (�111) for theformer and (002) for the latter.

Cross-section electron microscopy showed that, irre-spective of the substrate material, all oxide films in thiswork demonstrated columnar crystal morphology. Weobserved the same characteristically remarkable enhance-ment in the CuO (002)/(�111) in films deposited on vari-ous substrates such as glass slides, single-crystal siliconwafers, and glass slides coated with other oxide films suchas SnO2, TiO2 and ZnO, understandably owing to lack ofeffective seeding effect. Fig. 3 shows typical columnar struc-tures for CuO films deposited on Si (with its surface beingcovered by a thin layer of native amorphous silica �2 nmthick). In spite of the different oxygen percentages in thesputtering gas, both films exhibit similar columnar struc-tural features. The formation of columnar structureresulted in the formation of fibrous textures in the oxidefilms, thus leading to diffraction patterns dominated by

Fig. 3. Typical cross-section TEM images of CuO thin films: (a) 13%O2 in sputtering gas; (b) 77% O2 in sputtering gas. Numbered areasindicate positions for EELS analysis (see Fig. 5).

one major peak and the disappearance of the other peaks.Such columnar structure across the film thickness is consid-ered advantageous for PV devices, through the avoidanceof lateral grain boundary scavenging of photo-induced elec-tron and holes.

High-resolution transmission electron microscopy(HRTEM) was carried out to investigate the fine structuresof the materials. Fig. 4 shows a HRTEM image taken at the[�1–10] zone axis of the CuO crystal in the film depositedwith a low O2 percentage in the sputtering gas (13%). Theinset is the corresponding fast Fourier transform (FFT)image, showing diffraction spots contributing to theHRTEM image. The direction of the film normal is shownas a nearly horizontal arrow, which corresponds to the nor-mal direction of the CuO (�111) plane. This is consistentwith the XRD results that the (�111) is the major peakin the XRD pattern of the CuO films with lower oxygencontents in the sputtering gas.

The chemical states of Cu were studied using electronenergy loss spectroscopy (EELS). Fig. 5 shows representa-tive results from low- and as well as high-oxygen films. Theoverall characteristics of the Cu L edge in the films weredefined by EELS spectra from an area �200 nm in diame-ter, and spot collections corresponded to slit apertures�20 nm in diameter, as shown in the corresponding analy-sis regions in Fig. 3. It can be seen from Fig. 5a that theoverall Cu L edge is representative of that from spot 1,which is nearly the same as the standard Cu L edge fromCuO (consisting of a major peak and a minor shoulderon its right). Radically different spectra such as that fromspot 2 were observed occasionally, which is attributed tothe existence of the high-Cu phase Cu64O. The fact thatthe overall feature of the Cu L edge is dictated by that ofthe CuO phase is consistent with the XRD results whereinonly trivial evidence for the presence of Cu64O wasdetected.

Fig. 5b shows the L edge spectra from a high-oxygensample. The overall L edge is represented by those fromeach columnar grain of CuO. Interestingly, we found thatthe L edge from spots containing columnar boundaries isslightly different, with the secondary shoulder slightly

Fig. 4. HRTEM image of the low-oxygen thin sample deposited with13% oxygen in the sputtering gas. Inset shows the corresponding FFT.Arrow indicates normal direction of the film.

925 930 935 940 945 950 955 960

Spot 1

Background(a)

Spot 2

Energy-loss (eV)

925 930 935 940 945 950 955 960

Background(b)

Spot 3

Spot 2

Spot 1

Energy- loss (eV)

Fig. 5. EELS Cu L-edge from CuO thin films: (a) 13% O2 in sputteringgas; (b) 77% O2 in sputtering gas. Corresponding spot positions areshown in Fig. 3. Each spot size is 20 nm in diameter and thebackground size is 200 nm in diameter.

965 960 955 950 945 940 935 930 925

Cu 2p1/2

33%47%

77%

13%

CuO

Binding Energy (eV)

Cu 2p3/2

Fig. 6. Cu 2p spectra of CuO thin films, shown against oxygen contentin the sputtering gas balanced with Ar. The spectrum from a standardpristine CuO sample is shown for comparison.

126 Y. Shen et al. / Acta Materialia 85 (2015) 122–131

enhanced comparable to that from the low oxygen sample.It is worth noticing that such enhancement in the peak atthe minor shoulder is related to the local enrichment ofCu, so that in an extreme case, the peak intensities in theCu L edge were swamped by that from the Cu64O phase(Fig. 5a).

3.2. XPS analysis

Cu 2p XPS spectra of different CuO thin films are pre-sented in Fig. 6, with each showing the fingerprint featuresof characteristic binding peaks and associated satellite onesfor the CuO. It is noted that each spectrum is nearly thesame as that of pristine CuO powder, which suggests thatthe monoclinic CuO phase dictates the microstructures ofall of the thin films deposited with different levels of oxygenin the sputtering gas. The presence of a series of satellitepeaks is characteristic of CuO which has a d9 configurationin the ground state [42–45]. Furthermore, the energetic dif-ference (DE) between the ECu2p3/2 and ECu2p1/2 peaks is20 ± 0.1 eV, which is in good agreement with literaturedata.

Fine details in the high-resolution Cu 2p3/2 peak andneighbouring satellites for representative CuO films are

demonstrated with peak deconvolution in Fig. 7. Startingfrom the standard spectrum from the pristine CuO powder,both the binding and satellites were fitted to two peaks. Thepeak position at 933.4 eV is attributed to the covalent partof Cu2+binding in the CuO phase [38,42,44], and the otherpeak at 934.9 eV is due to the ionic Cu2+binding [45,46].This is in excellent agreement with well-documented litera-ture reports that there exist two kinds of Cu2+ binding inpristine CuO. The reported percentage of covalent Cu–Obonding is 52%, and the rest is ionic. It is reasonable thatthe broad satellite peak next to the binding peak can alsobe attributed to two peaks, due to its association with dif-ferent types of Cu–O bonding, which results from the d9

characteristics for CuO bonding wherein Cu contributes ad electron from the completed d-shell for d–p hybridizationwith O orbitals. On the other hand, all the d electrons forCu in the cuprous oxide (Cu2O) remain in the completedd-shell and behave like pseudo-core electrons, leaving Cu–O bonding solely to the s electrons from Cu, thus leadingto a simpler XPS spectrum without any satellites. Thereported Cu+ 2p3/2 peak lies in the range 932.2–932.5 eV[47–49], which is apparently not present in Fig. 7a.

When CuO thin films are deposited with low oxygenfractions in the sputtering gas, an extra minor peak canbe distinguished in the Cu 2p3/2 peak, e.g. Fig. 7b and cfor oxygen fractions of 13% and 33%, respectively. Theextra peak at 932.5 or 932.6 eV lies in the range for Cu withlower valency, from Cu+ to Cu0 [47–49]. It is thereforeimpossible to distinguish the valence states for any highCu oxides with valencies between +1 and 0, because thebinding energies are very close, within 0.1 eV [42]. It istherefore useful to consult the information obtained fromthe Auger transition. It was reported that the positions ofthe L3VV Auger transition in XPS spectra are 569.6 and571.9 eV for Cu+ and Cu0 respectively [50]. Fig. 8 showsthe L3VV Auger spectra of CuO thin films deposited withdifferent oxygen fractions in the sputtering gas. It is appar-ent that there is no noticeable evidence for Cu0 (expected at571.9 eV), and the secondary shoulder on the right(932.5 eV) cannot offer decisive insight, as it is present evenin the spectrum from pristine CuO phase. Therefore, a

948 945 942 939 936 933 930 927Binding Energy (eV)

Cu2p3/2

932.5eV

933.6eV935.2eV

940.7eV943.2eV

(b) O2 in gas: 13%

948 945 942 939 936 933 930 927Binding Energy (eV)

Cu2p3/2

933.4eV

934.9eV

943.3eV940.8eV

(a) CuO

948 945 942 939 936 933 930 927Binding Energy (eV)

Cu2p3/2

933.5eV

935.6eV

940.8eV943.4eV

(d) O2 in gas: 47%

948 945 942 939 936 933 930 927Binding Energy (eV)

933.1eV

934.7eV

940.85eV943.4eV

Cu2p3/2(e) O2 in gas: 77%

948 945 942 939 936 933 930 927Binding Energy (eV)

Cu2p3/2

932.6eV

933.7eV

935.0eV

940.9eV943.5eV

(c) O2 in gas: 33%

Fig. 7. Peak deconvolution for Cu 2p3/2 in: (a) pure CuO powder; (b) 13% O2 in sputtering gas; (c) 33% O2 in sputtering gas; (d) 47% O2 in sputteringgas; (e) 77% O2 in sputtering gas.

Y. Shen et al. / Acta Materialia 85 (2015) 122–131 127

self-consistent conclusion can only be drawn by consideringthe XRD and TEM/EELS results, such that the high-cop-per oxide is attributed to the Cu64O phase. It is worthemphasizing that only minor amounts of such a high-cop-per phase were present in thin films deposited with <40%O2 in the sputtering gas (region I in Fig. 2). The low-valenceCu in high-copper oxide completely disappeared when theoxygen fraction was >40% in the sputtering gas, and a com-positional plateau was maintained at about the CuO stoi-chiometry with oxygen fractions of between 45% and 60%in the sputtering gas. Positive deviation from the ideal stoi-chiometry in oxygen content in the thin films occurred withoxygen fractions of >60% in the sputtering gas. Also, noevident change has been observed in the valence band ofany of the CuO films studied in this work. Atypical valanceband spectrum of a CuO thin film (see Fig. 8b) shows anedge exactly at the instrumental Fermi level. This is typicalfor p-type semiconductors, for which no states exist abovethe Fermi level. Overall, the CuO phase dominates the

microstructures of all the sputter-deposited films of thiswork, and all CuO films exhibit p-type conductivity.

The high-resolution O 1s peaks from some representa-tive CuO thin films are compared with the standard peakfrom pristine CuO powder in Fig. 9. The spectrum fromthe pristine CuO powder is in excellent agreement withthe literature data [42], and this broad peak could be fittedinto three peaks located at 592.6, 531.3 and 533.0 eV. Themain peak with the lowest binding energy is attributed toCu2+ bonding, and according to Ref. [42], the minor peakat the highest energy of 533.0 eV corresponds to surfacehydroxide or chemisorbed oxygen. The shoulder peakbetween them is attributed to defective oxygen in the bulkmaterial. It is interesting to note that these three peaksare present in all of the sputter-deposited CuO films, withthe defective and surface oxygen peaks significantlyenhanced with respect to those from the pristine power.

In addition, the film deposited with a low-oxygen sput-tering gas (Fig. 9b) demonstrated an additional peak

30 25 20 15 10 5 0 -5Binding Energy (eV)

(a)

(b)

550 560 570 580 590Binding energy(eV)

Pure CuO powder

O2 fraction: 77%

O2 fraction:13%

Cu3LVV Cu2+

Cu+

Fig. 8. (a) CuL3VV Auger spectra in different CuO thin films, whereinthe positions for Cu2+ and Cu+ are indicated. No noticeable differenceexists between the spectra from CuO films and that from the standardCuO powder. (b) Typical valance band representative of all CuO filmsin this work.

128 Y. Shen et al. / Acta Materialia 85 (2015) 122–131

between the main and defective oxygen peaks, with a bind-ing energy of 530.0 eV. This is consistent with XRD andTEM/EELS analysis which showed that a high-copperoxide existed in such films. On the other hand, in the filmwith the highest oxygen sputtering gas, a fourth peakappeared at the highest energy of 534.3 eV, which is5.1 eV above the binding energy for the main peak(Fig. 9d). Such a high-energy shift indicates the presenceof physically absorbed gaseous oxygen on the film surface.

Overall, one can see from the current results that sput-tering deposition led to significantly enhanced defectiveand surface oxygen in the CuO films. This is reasonabledue to the fact that RF magnetron sputtering is known tosynthesize materials at non-equilibrium conditions, whereinthe films are bombarded by energetic depositing ions underlimited diffusion during condensation of sputtered ions.The highly clean surface from such a physical process alsooffers favourable sites for surface contamination from theenvironment. This was confirmed by a short period ofin situ ion etching (Ar+) within the XPS instrument, thoughion etching tends to reduce metal oxides and one needs tobe cautious in over-etching oxides for XPS analysis [41].

3.3. Electronic/electrical properties

The optical band gaps of CuO thin films were also mea-sured using the standard a1/2/hv relationship, to obtain anindirect gap �1.41 ± 0.03 eV for all films in regions II andIII in Fig. 2. This shows that once the films were made ofonly CuO, the optical gap was independent of the fractionof oxygen in the sputtering gas.

Fig. 10 shows the Hall properties of CuO thin films fab-ricated using different oxygen fractions in the sputteringgas. All thin films showed a p-type characteristic for electri-cal conductivity, which is consistent with the observed XPSvalence bands exhibiting the same position of the top of thevalence bands at the Fermi level. It can be seen from Fig. 10that a higher hole concentration corresponds to a lowermobility. Referring back to Fig. 2, it can be seen that thenear-stoichiometric CuO films in region II demonstratedthe lowest carrier concentrations (of the order of1015 cm�3) and highest carrier mobilities (2–6 cm2V�1s�1).It is not suprising that higher hole mobility correspondedto lower carrier concentration, owing to reduced scatteringof holes at the native defects (as self-dopants). The bestmobility for CuO films of this work is lower than the holemobility in sputter-deposited Cu2O films with a hole con-centration about an order of magnitude higher(20 cm2V�1s�1 [51]).

According to first-principles calculations in the frame-work of density functional theory, the energeticallyfavoured defects in CuO are Cu vacancies (VCu) [52], whichinduce shallow acceptor states at the top of the valenceband. The existence of copper vacancies leads to an enrich-ment of oxygen content. On the other hand, in oxygen-defi-cient materials, interstitial copper atoms or oxygenvacancies are expected. As CuO is an ionic compound,charge transfer from the metal sites to the oxygen sites isbehind the p- or n-type electrical conductivity. Therefore,the existence of copper vacancies results in more oxygenions, which accept electrons, thus exhibiting p-type conduc-tivity. The existence of copper interstitials or oxygen vacan-cies, however, leads to more electrons being donated frommetal to oxygen sites, and hence the associated n-type con-ductivity. However, their high energies of formation maketheir presence in CuO rather difficult. The fact that p-typecharacteristics were observed in all the CuO films in thiswork thus suggests that all of the CuO phase in the filmsof this work had at least the minimal oxygen content closeto the stoichiometry. This is consistent with the EELSobservation that all CuO phases in low- or high-oxygensputtered thin films had nearly the same features in theirCu 2p spectra (Fig. 5). In the low-oxygen films (region Iin Fig. 2) high-copper phase(s) such as Cu64O appeared.It seems that the higher carrier concentration in region Iis associated with the presence of such high-copperphase(s). From region II to region III, increasing oxygencontent led to an increasing amount of copper vacancies,thus increasing the carrier concentration. Finally, whenthe oxygen fraction in the sputtering gas was >60%, the car-rier concentration of CuO thin films tended to reach aplateau.

The fact that all the films in this work exhibited almostthe same band gap is consistent with the microstructuralobservation that CuO was the major phase in all the films.The deviation of oxygen content from the ideal CuOstoichiometry contributed to causing self-doping in the

524 526 528 530 532 534 536 538

O1s

Binding Energy (eV)

(a) CuO

529.6eV

531.3eV

533.0eV

524 526 528 530 532 534 536 538Binding Energy (eV)

O1s(b) O2 in gas: 13%

529.3eV

530eV

531.3eV

532.8eV

524 526 528 530 532 534 536 538Binding Energy (eV)

O1s(c) O2 in gas: 47%

529.5eV 531.4eV

532.9eV

524 526 528 530 532 534 536 538

O1s(d) O2 in gas: 77%

Binding Energy (eV)

529.2eV 531eV

532.7eV

534.3eV

Fig. 9. Examples for peak deconvolution of O1s in: (a) pristine CuO powder; (b) 13% O2 in sputtering gas; (c) 47% O2 in sputtering gas; (d) 77% O2 insputtering gas.

0 10 20 30 40 50 60 70 80

-1

0

1

2

3

4

5

6

7

Car

rier c

once

ntr.

(1016

cm-3

)

-1

0

1

2

3

4

5

6

7

O2% in gas

Mobility (cm

2V-1s

-1)

Fig. 10. Hall properties of CuO thin films deposited with differentpercentages of oxygen in sputtering gas.

Y. Shen et al. / Acta Materialia 85 (2015) 122–131 129

materials, so that they exhibited p-type conductivity. Suchself-doping via native defects (VCu) in CuO involvesshallow acceptors [52], which could not induce any evidentchange in the band gap of the materials.

The I–V characteristics of CuO thin films depositedusing different oxygen fractions in the sputtering gas weremeasured with a typical two-probe method with coppercontacts being deposited on the top of the films. The resultsare summarized in Fig. 11. For the thin films of low carrierconcentrations (region II in Fig. 2), e.g. CuO film depositedusing a sputtering gas containing 47% oxygen, the I–V

curve showed a double Schottky feature. On the otherhand, the I–V correlations were linear for films of high car-rier concentrations (in either region I or region III inFig. 2). This suggests that, fundamentally, the junctionbetween Cu and CuO is Schottky in nature. The ohmic I–V feature in films with high carrier concentrations wasattributed to the narrowed space charge region at the junc-tion, thus resulting in a tunnelling effect for hole extractionfrom the semiconductor to the electrodes.

The work function of p-type CuO is �5.2–5.3 eV [53,54],which is higher than that of Cu (4.5–5.1 eV) [55]. The Scho-ttky barrier between a metal and a semiconductor is con-trolled by the work function of the metal and the electronaffinity of the semiconductor, so that for a p-type materialthe barrier according to the Anderson model is:

uB ¼ Eg=qþ v� um; ð1Þwhere Eg is the band gap, q is the elementary charge, qv isthe electron affinity of the semiconductor and qum definesthe work function of the metal contact. According to thereported data from the literature, the Schottky barrier iscalculated to be 1.5–2.3 eV using the Anderson equation.The actual Schottky barrier is usually much lower thanthe Anderson value, due to charge screening and the pres-ence of defects at the junction. It is worth pointing out thatthe Schottky junctions of this work were quite stable andno decay of the I–V characteristics was observed after pro-longed storage at ambient temperature.

The ability to control the hole concentration in CuOenables fine-tuning of the width of space charge regionsin devices based on either a p–n junction, or a Schottky

-1.0 -0.5 0.0 0.5 1.0-6

-3

0

3

6

71%

63%

57%

Voltage (V)

33%

(b)

-1.0 -0.5 0.0 0.5 1.0

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0C

urre

nt (

μA)

Cur

rent

(μA

)

Voltage (V)

O2 in gas: 47%(a)

Fig. 11. I–V characteristics of CuO thin films deposited at differentpercentages of oxygen in sputtering gas: (a) 47%; (b) low and highO2%.

130 Y. Shen et al. / Acta Materialia 85 (2015) 122–131

junction, so that harvesting of solar energy could be real-ized by using a low-cost materials system based on CuO,which is environmentally friendly and an abundantresource. This would enable production of high-efficiencysolar cells based solely on oxides [28,29]. Work is underwayto fabricate oxide-based solar cells making use of CuO asthe optical absorber layer.

4. Conclusions

CuO thin films were deposited using reactive RF sput-tering deposition, and the oxygen content could be tunedby varying the amount of oxygen in the sputtering gas.While a minor amount of high-copper oxide was observedin films deposited with low-oxygen sputtering gas, filmscontaining only the CuO phase were synthesized using anAr + O2 sputtering gas containing sufficient oxygen.

All CuO films were p-type semiconductors, with theacceptor states being attributed to the existence of coppervacancies. Low hole concentration and high hole mobilitywere achieved in films of compositions close to the idealstoichiometry. Both positive and negative deviation of oxy-gen content from the ideal stoichiometry was found toincrease hole concentration and decrease hole mobility.

A Schottky barrier exists between the Cu/CuO junctionwhen the hole concentration in the semiconductor was low(1015–1016 cm�3). High hole concentrations, however, ledto tunnelling at the metal–semiconductor junction, thusleading to ohmic I–V characteristics. The ability to controlthe hole concentration enables tuning of the I–V character-istics, which are essential in using CuO films in photonicdevices such as low-cost PV cells, either in the form of aSchottky diode or a p–n junction diode, using an environ-mentally friendly and abundant oxide.

Acknowledgments

The present work is partially supported Technology StrategyBoard of the UK and by the Natural Science Foundation of China(Nos. 51001091, 111174256, 91233101) and the FundamentalResearch Program from the Ministry of Science and Technologyof China (No. 2014CB931704). Equipment access to the LeedsESPRC Nanoscience and Nanotechnology Facility (LENNF) isalso acknowledged.

References

[1] M. Engin, F. Atay, S. Kose, V. Bilgin, I. Akyuz, J. Electron.Mater. 38 (2009) 787.

[2] J.H. Hsieh, P.W. Kuo, K.C. Peng, S.J. Lin, J.D. Hsueh, S.C.Chang, Thin Solid Films 516 (2008) 5449.

[3] H. Yu, J. Yu, S. Liu, S. Mann, Chem. Mater. 19 (2007) 4327.[4] F. Illas, D. Munoz, Coen de Graaf, Iberio de P.R. Moreira,

Chem. Phys. Lett. 379 (2003) 291–296.[5] E. Morenzoni, B.M. Wojek, A. Suter, T. Prokscha, G.

Logvenov, I. Bozovic, Nat. Commun. 2 (2012) 272.[6] A. Mittiga, E. Salza, F. Sarto, M. Tucci, R. Vasanthi, Appl.

Phys. Lett. 88 (2006) 163502.[7] K. Cheng, Q. Li, J. Meng, X. Han, Y. Wu, S. Wang, et al.,

Sol. Energ. Mat. Sol. C 116 (2013) 120.[8] T.J. Hsueh, C.L. Hsu, S.J. Chang, P.W. Guo, J.H. Hsieh, I.C.

Chen, Scr. Mater. 57 (2007) 53.[9] P. Raksa, S. Nilphai, A. Gardchareon, S. Choopun, Thin

Solid Films 517 (2009) 4741.[10] J.F. Pierson, A. Thobor.-Keck, A. Billard, Appl. Surf. Sci.

210 (2003) 359.[11] A.A. Ogwu, E. Bouquerel, O. Ademosu, S. Moh, E. Crossan,

F. Placido, Acta Mater. 53 (2005) 5151.[12] W.T. Yao, S.H. Yu, Y. Zhou, J. Jiang, Q.S. Wu, L. Zhang, J.

Jiang, J. Phys. Chem. B 109 (2005) 14011.[13] F. Marabelli, G.B. Parravicini, F. Salghetti-Drioli, Phys. Rev.

B 52 (1995) 1433.[14] F. Marabelli, G.B. Parravicini, Physica B 199&200 (1994)

255.[15] L.A. Patil, D.R. Patil, Sens. Actuators B: Chem. 120 (2006)

316.[16] Y.Z. Feng, X.L. Zheng, Nano Lett. 10 (2010) 4762.[17] Y.W. Zhu, T. Yu, F.C. Cheong, X.J. Xu, C.T. Lim, V.B.C.

Tan, J.T.L. Thong, C.H. Sow, Nanotechnology 16 (2005) 88.[18] C.T. Heieh, J.M. Chen, H.H. Lin, H.C. shih, Appl. Phys.

Lett. 83 (2003) 3383.[19] F. Gao, X.J. Liu, J.S. Zhang, M.Z. Song, N. Li, J. Appl.

Phys. 111 (2012) 084507.[20] C.C. Chusuei, M.A. Brookshier, D.W. Goodman, Langmuir

15 (1999) 2806.[21] J. Zhang, J. Liu, Q. Peng, X. Wang, Y. Li, Chem. Mater. 18

(2006) 867.[22] X. Zhang, G. Wang, X. Liu, J. Wu, M. Li, J. Gu, et al., J.

Phys. Chem. C 112 (2008) 16845.[23] S.Y. Sung, S.Y. Kim, K.M. Jo, J.H. Lee, J.J. Kim, S.G. Kim,

K.H. Chai, S.J. Pearton, D.P. Norton, Y.W. Heo, Appl.Phys. Lett. 97 (2010) 222109.

Y. Shen et al. / Acta Materialia 85 (2015) 122–131 131

[24] C. Dıaz-Guerra, M. Vila, J. Piqueras, Appl. Phys. Lett. 96(2010) 193105.

[25] G. Conibeer, Mater. Today 10 (2007) 42.[26] N. Lopez, L.A. Reichertz, K.M. Yu, K. Campman, W.

Walukiewicz, Phys. Rev. Lett. 106 (2011) 028701.[27] A. Polman, H.A. Atwater, Nat. Mater. 11 (2012) 174.[28] L. Zhu, G. Shao, J.K. Luo, Semicond. Sci. Tech. 26 (2011)

085026.[29] L. Zhu, G. Shao, J.K. Luo, Solid State Sci. 14 (2012) 857–863.[30] L. Zhu, J.K. Luo, G. Shao, W.I. Milne, Sol. Energ. Mat. Sol.

C 111 (2013) 141.[31] C. Wadia, A.P. Alivisatos, D.M. Kammen, Environ. Sci.

Technol. 43 (2009) 2070.[32] G. Shao, J. Phys. Chem. C 112 (2008) 18677.[33] G. Shao, J. Phys. Chem. C 113 (2009) 6800.[34] X. Han, G. Shao, J. Phys. Chem. C 115 (2011) 8274.[35] X. Han, K. Song, L. Lu, Q. Deng, X. Xia, G. Shao, J. Mater.

Chem. C 1 (2013) 3736.[36] X. Han, G. Shao, Phys. Chem. Chem. Phys. 15 (2013) 9581.[37] L. Debbichi, M.C. Marco de Lucas, J.F. Pierson, P. Kruger,

J. Phys. Chem. C 116 (2012) 10232.[38] M. Hari Parsad Reddy, J.F. Pierson, S. Uthanna, Phys.

Status Solidi A 209 (2012) 1279.[39] M.T.S. Nair, L. Guerrero, O.L. Arenas, P.K. Nair, Appl.

Surf. Sci. 150 (1999) 143.[40] M.Y. Lin, C.Y. Lee, S.C. Shiu, I.J. Wang, J.Y. Sun, W.H.

Wu, Y.H. Lin, J.S. Huang, C.F. Lin, Org. Electr. 11 (2010)1828.

[41] X.H. Xia, L. Lu, A.S. Walton, M. Ward, X.P. Han, R.Brydson, J.K. Luo, G. Shao, Acta Mater. 60 (2012) 1974.

[42] J. Ghijsen, L.H. Tjeng, J. van Elp, H. Eskes, J. Westerink,G.A. Sawatzky, Phys. Rev. B. 38 (1988) 11322.

[43] T. Ghodselahi, M.A. Vesaghi, A. Shafiekhani, A. Baghizadeh,M. Lameii, Appl. Surf. Sci. 255 (2008) 2730.

[44] G. Panzner, B. Egert, H.P. Schmidt, Surf. Sci. 151 (1985) 400.[45] A.E. Palomares, F. Marquez, S. Valencia, A. Corma, J. Mol.

Catal. A Chem. 162 (2000) 175.[46] Z. Dang, J.F. Gu, L. Yu, Appl. Catal. 63 (1990) 259.[47] S. Poulston, P.M. Parlett, P. Stone, M. Bowker, Surf.

Interface Anal. 24 (1996) 811.[48] P. Liang, D. Prendergast, F. Borondics, S. Porsgaard, L.

Giovanetti, E. Pach, J. Newberg, H. Bluhm, F. Besenbacher,M. Salmeron, Chem. Phys. 138 (2013) 024704.

[49] I.J. Kartio, C.I. Basilio, R.H. Yoon, Langmuir 14 (1998)5274.

[50] D. Tahir, S. Tougaard, J. Phys. Condens. Matter. 24 (2012)175002.

[51] Z.Q. Yao, S.L. Liu, L. Zhang, B. He, A. Kumar, X. Jiang,W.J. Zhang, G. Shao, Appl. Phys. Lett. 101 (2012) 042114.

[52] D.X. Wu, Q. Zhang, Phys. Rev. B 73 (2006) 235206.[53] L. Liao, Z. Zhang, B. Yan, Z. Zheng, Q.L. Bao, T. Wu, C.M.

Li, Z.X. Shen, J.X. Zhang, H. Gong, J.C. Li, T. Yu,Nanotechnology 20 (2009) 085203.

[54] F.P. Koffyberg, F.A. Benko, J. Appl. Phys. 53 (1982) 1173.[55] CRC Handbook on Materials Chemistry, 2008, pp. 12–114.