Poole–Frenkel effect in sputter-deposited CuAlO2+x nanocrystals

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2013 Nanotechnology 24 165705

(http://iopscience.iop.org/0957-4484/24/16/165705)

Download details:

IP Address: 128.148.252.35

The article was downloaded on 09/09/2013 at 21:47

Please note that terms and conditions apply.

View the table of contents for this issue, or go to the journal homepage for more

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING NANOTECHNOLOGY

Nanotechnology 24 (2013) 165705 (7pp) doi:10.1088/0957-4484/24/16/165705

Poole–Frenkel effect in sputter-depositedCuAlO2+x nanocrystals

Arghya Narayan Banerjee and Sang Woo Joo

School of Mechanical Engineering, Yeungnam University, Gyeongsan 712-749, Korea

E-mail: banerjee arghya@hotmail.com and swjoo@yu.ac.kr

Received 20 December 2012, in final form 8 March 2013Published 27 March 2013Online at stacks.iop.org/Nano/24/165705

AbstractField-assisted thermionic emission within a sputter-deposited, nanocrystalline thin film ofCuAlO2.06 is observed for the first time, and explained in terms of the Poole–Frenkel model.The presence of adsorbed oxygen ions as trap-states at the grain boundary regions of thenanostructured thin film is considered to manifest this phenomenon. Under an applied field,the barrier of the trap potential is lowered and thermal emission of charge carriers takes placeat different sample temperatures to induce nonlinearity in the current (I)–voltage (V)characteristics of the nanomaterial. Fitting of the Poole–Frenkel model with the I–V datashows that the nonlinearity is effective above 50 V under the operating conditions.Calculations of the energy of the trap level, acceptor level and Fermi level reveal the existenceof deep level trap-states and a shallow acceptor level with acceptor concentration considerablyhigher than the trap-states. Hall measurements confirm the p-type semiconductivity of thefilm, with a hole concentration around 1018 cm−3. Thermopower measurements give aroom-temperature Seebeck coefficient around 130 µV K−1. This temperature-dependentconductivity enhancement within CuAlO2 nanomaterial may find interesting applications intransparent electronics and high-voltage applications for power supply networks.

(Some figures may appear in colour only in the online journal)

1. Introduction

Delafossite CuAlO2 and its doped-versions [1, 2] are some ofthe very important p-type semiconducting transparent oxide(p-STO) materials which have photovoltaic [3, 4], opto-electronic [5–7], excitonic [8–10], thermoelectric [11–13],magnetic [14], catalytic [15], photocatalytic [16], photo-electrochemical [17–19], field emission [20, 21], gas sens-ing [22, 23], superconducting [24] and self-cleaning [25]properties. Amongst these, the optoelectronic properties ofthis material have recently attracted renewed interest forpotential applications in so-called ‘invisible’ or ‘transparentelectronics’ [5, 26], where a combination of p-STOs with thealready established n-STOs (e.g. ZnO, ITO etc), in the formof a p–n junction, would generate electricity by the absorptionof the UV part of solar radiation, yet becomes transparentto visible solar light. Thus it can act as a ‘functional’window for UV absorption and energy generation. One of themost important issues related to CuAlO2-based transparent

diodes is that the junction efficiency is considerably affectedby the poor electrical conductivity (σ ) of CuAlO2 withrespect to commonly used n-STOs such as ZnO, ITO (andtheir doped-versions) etc. Although the optical properties arecomparable, the σ -value of (doped and undoped) CuAlO2is more than two orders of magnitude less than that ofthe common n-STOs. Therefore the improvement in theconductivity of this material is very important for its efficientuse in transparent electronics.

It is well known that the p-type conductivity of thismaterial is due to the excess (non-stoichiometric) oxygenintercalation within the material according to the followingdefect equilibrium [12, 27]:

O2(g) = 2OxO + V−Cu + V−3

Al + 4h+ (1)

where OO,VCu, VAl and h denote lattice oxygen, Cuvacancy, Al vacancy and hole, respectively. Superscripts x,− and + denote effective neutral, negative and positivecharge states, respectively. Deposition and annealing of this

10957-4484/13/165705+07$33.00 c© 2013 IOP Publishing Ltd Printed in the UK & the USA

Nanotechnology 24 (2013) 165705 A N Banerjee and S W Joo

material at excess oxygen atmosphere generally inducesnon-stoichiometric oxygen (within thermodynamical limit)into the material and enhances the hole conductivity [2].On the other hand, the hole transport phenomenon withinthis material is reported to follow the variable rangehopping mechanism (log σ ∼ T−1/4) at low temperatureand thermally activated small polaron (log σT ∼ T−1)

or semiconductor-type conduction (log σ ∼ T−1) at roomtemperature and above [28–30]. Ingram et al [29] inferredthe high-temperature conduction mechanism within their as-synthesized CuAlO2 polycrystals as small polaron conductionbecause of the manifestation of temperature-independentthermopower and very low thermally activated hole mobility.On the other hand, Yanagi et al [28] observed semiconductor-type hole conduction within their as-synthesized CuAlO2 thinfilm near room temperature, and speculated it to be due tothe relatively higher activation energy of the material. In ourprevious works, we have reported the temperature-dependentelectrical and thermoelectric properties of sputter-depositedCuAlO2 thin films, which follow the semiconductor-typethermally activated conduction mechanism [11, 27, 31–33],as the corresponding activation energy falls within a similarrange as reported by Yanagi et al [28]. In the current study,we have, for the first time (to best of our knowledge), reportedthe temperature-dependent non-Ohmic current–voltage (I–V)characteristics within our sputter-deposited nanocrystallineCuAlO2, which is attributed to the Poole–Frenkel effectof field-assisted thermionic emission from various trap-states present within the grain boundary regions of thenanostructured thin film. The I–V data are fitted withboth a Schottky model and a Poole–Frenkel model, andthe best fits show a transition from Schottky emission(near electrode contacts) at lower fields to Poole–Frenkelemission (within the sample volume) at higher fields. Theestimated transition region between these two regimes iscorrelated with the Fermi energy, acceptor and trap-stateconcentrations to get an approximate band-energy picture ofthe nanomaterial. This temperature-dependent enhancementof the carrier concentration not only offers high-efficiencyapplications of this material in transparent electronics, but alsoopens up the possible uses in high-voltage applications forpower supply networks and interfacial electrical stress gradingmechanism at metal–insulator composites [34].

2. Experimental procedure

Nanocrystalline CuAlO2 thin film fabrication is a two-stepprocess of sintered powder preparation followed by thin filmsputter deposition. In the first step, a CuAlO2 sputteringtarget is fabricated by sintering a stoichiometric mixtureof Cu2O and Al2O3 (99.99%, Sigma Aldrich, Korea) in afurnace (Fisher Scientific, Korea) at 1100 ◦C for 24 h. Every6 h, the mixture is taken out of the furnace after propercooling, remixed and again placed into the furnace at thesame temperature. Next, the sintered body is reground andpressed into a pellet by hydrostatic pressure and then placedinto a grooved aluminum holder by appropriate arrangement,to be used as the target for sputtering. The direct-current (DC)

magnetron sputtering unit is a conventional vertical sputteringsystem, where the target is placed as the upper electrodeand connected to the negative terminal of a high-tensionpower supply, and the lower electrode (with an electrodedistance of 2.0 cm) is used as the substrate holder andgrounded. The film deposition is performed on Si substrates(Fisher Scientific, Korea). Before placing into the depositionchamber, the Si substrates are first immersed in 20% HFsolution for 5 min to remove the native surface oxidelayers and then ultrasonically cleaned by standard substratecleaning procedures using de-ionized water, isopropyl alcoholand piranha stripper (Duskan Pure Chemicals, Korea). Inthe second step, the sputter deposition is performed in anO2-diluted Ar (2:3 volume ratio) atmosphere. Initially, thesputtering chamber is evacuated by differential pumpingarrangements to a base pressure of 10−6 mbar, followedby pre-sputtering of the target for 5 min to remove surfacecontamination (if any) and then the shutter is displaced toexpose the substrates to the sputtering plasma. The depositionis performed at a sputtering pressure/voltage/current (density)of 0.2 mbar/1.2 kV/10.0 mA cm−2, respectively, witha substrate temperature of 250 ◦C. The deposition timeis taken around 30 min. The lower deposition time ischosen mainly to reduce the particle agglomeration andmaintain the nanostructure of the CuAlO2 thin film. Aftersputtering, the as-synthesized thin film is annealed in thesame deposition chamber (without breaking vacuum) underO2-atmosphere (0.2 mbar) to enhance hole conductivity witha post-annealing temperature/time around 350 ◦C/30 min,respectively. The details of the deposition procedures aredescribed elsewhere [8].

3. Results and discussions

The structural characterization of the nanocrystalline thinfilm is performed in a powder x-ray diffractometer (XRD,Philips XPertPowder) in θ–2θ mode with an x-ray wavelengthof 1.5405 A. Figure 1 represents the XRD data of ananocrystalline CuAlO2 thin film deposited on a Si substratewith background correction. The inset of figure 1 shows thecorresponding XRD graph for the sintered sputter target. Arelative comparison (cf table 1) of the d-values of the sinteredpowder (dpowder) and nanocrystalline thin film (dthin-film) withthe Joint Committee on Powder Diffraction Standards—TheInternational Center for Diffraction Data (JCPDS-ICDD) FileCard (dJCPDS) [35] reveals proper formation of phase-puredelafossite CuAlO2 with a rhombohedral crystal structurebelonging to the R3m space group.

The microstructural analysis of the nanocrystallineCuAlO2 thin film is performed with a scanning transmissionelectron microscope (STEM, Tecnai G2 F20). For high-resolution (HR) TEM characterizations, the as-depositednanoparticles are lifted off from the substrates and dispersedin alcohol, followed by stirring and drop-casting onto thecarbon-coated copper grids. Thereafter, these grids are driedproperly and introduced into the TEM chamber for imaging.Figure 2(a) shows the low-magnification TEM image of thenanocrystalline thin film. The size distribution (shown in

2

Nanotechnology 24 (2013) 165705 A N Banerjee and S W Joo

Figure 1. XRD graph of nanocrystalline CuAlO2 thin film. Inset:XRD graph of the corresponding sintered sputter target.

Table 1. Comparison of d-values of CuAlO2 target andnanocrystalline film with JCPDS-ICDD file card.

h k ldpowder

(A)dthin-film-XRD(A)

dthin-film-SAED(A)

dJCPDS [35](A)

0 0 6 2.82 2.80 — 2.831 0 1 2.45 2.49 2.47 2.450 1 2 2.37 2.35 — 2.381 0 4 2.13 2.11 2.12 2.141 0 7 1.73 1.73 – 1.730 1 8 1.61 1.60 1.62 1.61

the inset) shows an average particle size around 25 nm.The particle size is calculated by standard image analysissoftware and each particle diameter is calculated along twomutually perpendicular directions and then averaged out.Figure 2(b) shows the HRTEM image of a single CuAlO2nanocrystal of size around 20 nm. The micrograph clearlyshows the characteristic growth direction of the crystalplanes with proper lattice spacings. The inset representsthe corresponding selected-area electron diffraction (SAED)pattern, which corresponds to the characteristic crystal planesof nanocrystalline CuAlO2. Calculations of the correspondingd-values of the diffraction planes from the SAED micrograph,using a known camera constant, are inserted in table 1as dthin-film-SAED for comparison. The values match fairlywell with the JCPDS-ICDD file card [35]. The averagefilm thickness, obtained from cross-sectional field emissionscanning electron microscope (FESEM, Hitachi S-4200), isfound to be around 100–120 nm. Also the compositionalanalysis by energy dispersive x-ray (EDX) detector, equippedwith the STEM, shows nearly 3.0 at.% excess oxygen withinthe nanocrystalline thin film, with a Cu/Al ratio of almost 1.0(within the experimental limit). Therefore, the actual chemicalformula of the material can be given as CuAlO2.06 over thestoichiometric value.

The conductivity measurements of the nanocrystallinethin film are performed by a linear four-probe method under

Figure 2. (a) HRTEM micrograph of nanocrystalline CuAlO2 thinfilm. Inset represents the size distribution of the nanoparticles,(b) HRTEM micrograph of a single CuAlO2 nanocrystal. Insetshows the corresponding SAED micrograph.

a vacuum of 10−3 mbar and at a bias voltage of 20 V. Thecurrent–voltage (I–V) characterizations are done by a standardtwo-probe method since equation (3) (shown later) is modeledfor two-terminal system. All the contacts are made with Agpaint, which show Ohmic nature over a wide range of appliedvoltages. For Ohmic contacts, at low temperature (when thethermionic emission is negligible), the I–V characteristicsshould show linear behavior over the operating voltage range.In our case, for the verification of the Ohmic nature ofthe contacts, we have performed the I–V characterization ofthe samples at liquid nitrogen temperature (∼77 K), whichshowed linear behavior over the operating voltage range(0–200 V), indicating the Ohmic nature of the contacts.

The temperature variation (from room temperature, 30 ◦Cto 250 ◦C) of the conductivity (σ ) of the nanostructuredfilm is shown in figure 3. The Arrhenius plot showsthe semiconductor-type thermally activated conduction asreported previously by us and others for bulk CuAlO2film [1, 27, 28, 36]. The room-temperature conductivity

3

Nanotechnology 24 (2013) 165705 A N Banerjee and S W Joo

Figure 3. Temperature variation of conductivity of nanocrystallineCuAlO2 thin film. The error bars to the data points indicate that thevariations are within ±10% of the average data (indicated by theblue dashed lines against the solid black line).

(σRT) of the nanocrystalline thin film is found to be around0.20 S cm−1. Also the activation energy (Ea) is calculatedfrom the slope of the graph as 230 meV, which is comparableto the previously reported values [1, 11, 28]. Basically fora p-type semiconductor, Ea corresponds to the minimumenergy required to transfer holes from the acceptor levelto the valence band of CuAlO2. Apparently, the excessoxygen within CuAlO2.06 nanocrystalline film manifests thehole conductivity in accordance with the defect equilibriumshown in equation (1). To verify the p-type conductivityand to determine the carrier concentration, room-temperatureHall measurements are performed by the standard van derPauw method, with a rectangular van der Pauw configuration,where the electrical connections are made at the fourcorners of a square sample. The Hall coefficient of thematerial is found to be around 6.24 cm−3 C−1 with ahole concentration around 1.0 × 1018 cm−3. Clearly, therelatively lower room-temperature conductivity against thehole concentration is manifested by grain boundary scatteringwithin the nanocrystalline thin film that limits the electricalconductivity.

Now, it is well known that grain boundary barriersin polycrystalline ceramics can be reduced by increasingthe temperature of the sample [37]. To verify this, I–Vcharacterizations of the film at various constant sampletemperatures are performed under vacuum (10−3 mbar).Figure 4(a) represents the I–V characteristics of nanocrys-talline CuAlO2.06 thin film at different constant temperatures(which vary from 303 to 423 K). The curves show nonlinearbehavior, which is more profound at higher temperatures.Also with the increase in the voltage, the far-Ohmic behaviorbecomes more pronounced. This shape of the I–V curveis typical of field-enhanced thermionic emission over thebarrier, explained by the Poole–Frenkel model. Figure 4(b)shows the test fitting of the curves for the Poole–Frenkelmodel under our experimental conditions. The graphs depict

Figure 4. (a) I–V characteristics of nanocrystalline CuAlO2 thinfilm at different sample temperatures. (b) Test fitting of the I–V datafor the Poole–Frenkel effect. Inset: schematic representation of thePoole–Frenkel mechanism. (c) Plots of ln(J/ART2) versus V1/2 atdifferent sample temperatures.

that the nonlinear behavior starts around 50 V (shown bythe vertical dashed line in figure 4(b)). Previously, variousdoped-STOs (like F/Sb-doped SnO2, F-doped CdO, Sb-dopedCuAlO2 [38–41] etc) have been reported to show similarnonlinear I–V characteristics, which are explained in termsof Poole–Frenkel or space-charge-limited conduction mech-anisms via trap-states of dopant atoms at grain boundaries.In our case, we have observed this behavior for undoped

4

Nanotechnology 24 (2013) 165705 A N Banerjee and S W Joo

nanocrystalline CuAlO2.06 thin film and speculate that thepresence of adsorbed oxygen ions in the grain boundaryregions acts as the defect states and trap carriers to producepotential barrier (Et) at the grain boundaries (black lines inthe inset of figure 4(b)). Under the applied field (F), the barrierheight is reduced by a factor 1Et (with respect to the top ofthe valence band, Ev) in the direction of the applied field (redlines shown in the inset of figure 4(b)), and can be calculatedfrom the following relation as [42]:

1Et =

(q3F

πεr

)1/2

(2)

where q is the charge of the carrier and εr is the relativepermittivity of the material. Due to this lowering of thebarriers, thermal emission of charge carriers from thetrap-states takes place under the operating temperature, andthus increases the conductivity through the grain boundaryand also induces the nonlinearity in the I–V curves.

Mathematical expression for the Poole–Frenkel model ofcurrent emission can be given as (choosing e2

= 14.4 eV A asa conventional unit) [43]:

J = ART2 exp

[1

kBT

{(57.7 eV

εhfr d

)1/2

− Et

}](3)

where J is the current density for Poole–Frenkel emission,AR is the Richardson constant (=120 A cm−2 K−2), T isthe temperature in kelvin, kB is the Boltzmann constant, εhf

ris the high-frequency relative permittivity of the material, dis the electrode distance (in A) and Et is the barrier heightof the trap potential under zero field. According to the aboveequation, a graph between ln(J/ART2) versus V1/2 will followa linear curve, the intercept of which at zero field wouldgive the height of the trap potential barrier. Figure 4(c)represents the ln(J/ART2) versus V1/2 curves at four differentsample temperatures. From the figure it is clearly revealedthat all the curves have two distinct slopes. Now it iswell known that the conduction mechanism in a thin filmfollows two processes, one is the sample-limited conductionand the other is the electrode-limited conduction [44]. Inthe sample-limited conduction mechanism, the electrode–filmcontact becomes Ohmic in nature and the current is controlledby the body resistance of the sample. Therefore, in thiscase, the major portion of the applied voltage is droppedacross the sample volume and very little voltage dropoccurs at the contact points. On the other hand, for theelectrode (or contact)-limited conduction, the electrode–filmcontact becomes non-Ohmic (or blocking) type and thecurrent is controlled by the contact resistance. Therefore,the major portion of the applied voltage is dropped at thecontacts and very little across the sample volume. Hencethe current density is dominated by Poole–Frenkel emissionfor sample-limited conduction and by Schottky emission forelectrode-limited conduction. In our case, we have verifiedthe Ohmic nature of our contacts and, therefore, we expectsample-limited conduction in our sample. However, at lowervoltages, the potential drop at the contact resistance (howeversmall it be) becomes comparable to the applied voltage and,

therefore, the contacts become blocking type and follow theelectrode-limited Schottky emission, since most of the voltageappears across the contact resistance and very little acrossthe volume of the sample. On the other hand, at highervoltage, the potential drop across the body resistance of thesample dominates and the conduction mechanism becomesPoole–Frenkel type. For example, when the applied voltageis 5 V or less, calculations from figures 3 to 4 show thatonly 20% (or less) of the applied voltage is dropped acrossthe body resistance of the sample between the contacts, andthe rest is dropped at the two contacts. On the other hand,when the applied voltage is higher than 100 V, more than60% of the applied voltage is dropped across the samplebody resistance, thus justifying our above arguments. Now,if we compare the Richardson–Schottky equation and thePoole–Frenkel equation (equation (3)), we see that the twoequations are identical, except that the coefficient of V1/2

within the exponential term is reduced by a factor of 2 forthe Schottky equation [43]. That is why the ln(J/ART2)

versus V1/2 curves (at constant temperatures) follow twodistinct slopes for the two types of conductions. At lowervoltage, Schottky emission dominates, and with an increasein the voltage, Poole–Frenkel emission starts dominating;hence, the slopes of the curves deviate from higher to lowervalues, thus indicating a transition from electrode-limited tosample-limited conduction. In our case, we have observed thistransition roughly around 15–20 V (cf figure 4(c)). Similartypes of transition from electrode-limited to sample-limitedconduction has been observed in different oxide ceramics andorganic semiconductor films by various groups [44–46].

To determine the approximate trap energy levels (Et) inour nanomaterial, we have extrapolated the linear portionof the ln(J/ART2) versus V1/2 curves in the Poole–Frenkelregime to the ln(J/ART2) axis at zero field. The trap energylevels are obtained around 600–800 meV (with respect to thetop of the valence band) for different sample temperatures (cffigure 4(c)). Physically, these values of Et are the energiesrequired to transfer trapped holes from the trap level to thevalence band of CuAlO2, and obviously, at higher sampletemperature, more deep level traps are activated because of thefact that the trap energy levels are an increasing function of thesample temperature (cf figure 4(c)). In a polycrystalline thinfilm, there will be a statistical variation of the barrier heightof the trap-states (within thermodynamic limit). At a certainapplied voltage, the field-assisted lowering of the barrier willbe more or less the same for all the trap-states. Therefore, theprobability of thermionic emission will depend on the depthof the trap potential well. Obviously, at higher temperatures,probability of thermionic emission from deeper traps will behigher, thus creating a strong temperature dependence of thetrap energy levels. A relative comparison of the trap levelswith the activation energy (230 meV, cf figure 3) of thenanomaterial reveals that the trap levels are placed deeperinto the bandgap of CuAlO2 than the acceptor level withrespect to the top of the valence band. This type of banddiagram predicts that the acceptor concentration (Na) in oursample is higher than the trapped charge density (Nt) [47]. Toverify this, a relative comparison of Na and Nt is performed

5

Nanotechnology 24 (2013) 165705 A N Banerjee and S W Joo

Figure 5. Temperature variation of the Seebeck coefficient ofnanocrystalline CuAlO2 thin film. Inset: approximate band-energydiagram of the nanomaterial (not to scale).

by calculating the Fermi energy (Ef) of our sample viathermopower measurements presented below.

The thermoelectric power measurement of our sampleis performed by calculating the Seebeck coefficients (S) asa function of temperature gradient across the sample. Thetemperature gradient is created by keeping one end of the filmin a cold-head (@ room temperature, 30 ◦C) and the otherend in a hot-head, which is varied from room temperatureto 180 ◦C. The thermo-emfs generated between the hot- andcold-heads of the sample, at different hot-head temperatures,are used to determine the temperature-dependent Seebeckcoefficients of the material. The entire system is kept undervacuum condition (103 mbar). The temperature variationof the Seebeck coefficients is plotted in figure 5. Theroom-temperature Seebeck coefficient (SRT) is found to bearound +130 µV K−1. This value is comparable to theprevious reports on bulk CuAlO2 polycrystals and thin filmsby us and others [1, 11, 12]. The positive values of theSeebeck coefficients further confirm the p-type conductivityin our samples. Now, the temperature variation of the Seebeckcoefficients is related to the Fermi energy (Ef) of the materialaccording to the following formula [48]:

S =kB

q

(η +

Ef

kT

)(4)

where kB is the Boltzmann constant, q is the charge carrier(which is the same as an electronic charge in our case),η = 5/2 − s (where s is a scattering constant given as τ =τ0e−s, and τ is the relaxation time for electron scatteringand τ0 is a constant, which is a function of temperature butindependent of the electronic charge). From equation (4) it isapparent that a plot of S versus 1/T will be a straight line ifthe values of η and Ef become independent of temperature.Generally, near room temperature, these values becomeslowly-varying functions of temperature, and therefore, canbe approximated as constants within a certain range aroundroom temperature [48]. Therefore, assuming Ef and η to be

constants near room temperature, this straight line behaviorwill be pronounced near room temperature and will deviateconsiderably at higher temperatures. Therefore, we havecalculated the value of Ef from the slope of the curve nearroom temperature, which comes out around 190 meV. Thisvalue is comparable to the previously reported values of bulkCuAlO2 polycrystals and thin films by us and others [11, 49].An approximate band-energy diagram (assuming flat-bandpotential) of our nanocrystalline CuAlO2 thin film is shownin the inset of figure 5, taking into consideration the relativemagnitudes of Ea (∼230 meV),Et (∼600–800 meV) andEf (∼190 meV), obtained from figures 3–5, respectively.As the bandgap (Eg) of this type of transparent oxidesemiconductor is higher than the energy of a blue photon(∼3.1 eV) [2], the band-energy picture reveals the presenceof deep trap levels and shallow acceptor levels with respectto the top of the valence band (Ev), having a Fermi energy(Ef) placed in between Ev and Ea. This type of band structuredepicts the typical non-degenerate semiconducting behaviorwith the acceptors not fully ionized [50]. This is also thereason for the manifestation of semiconductor-type thermallyactivated conduction in our sample with a continuous increasein the conductivity with increasing temperature (cf figure 3).

Now to determine the relative magnitudes of ionizedacceptor (Na) and occupied trap (Nt) center concentrations,the position of the Fermi level is located by equating Na andNt, with the assumption that trapping is effective at all times(which seems reasonable because of the relatively higherroom-temperature resistivity of our nanocrystalline thin film).According to our approximate band structure shown in theinset of figure 5, the calculations of the above argument givethe following expression for Ef as [47]:

Ef = Ea − kBT ln(

Na

Nt

). (5)

Now inserting the values of Ef and Ea obtained fromfigures 3 and 5, respectively, we see that the magnitude of theionized acceptor center concentration is almost one order ofmagnitude higher than that of the trapping centers (at 300 K).Further, if we also consider the non-ionized acceptor centerspresent in the sample at room temperature (i.e. holes are yetto be transferred to the valence band from these acceptorcenters at room temperature), then it can be concluded thatthe acceptor concentration is considerably higher than the trapcharge density and agrees well with the Poole–Frenkel modeldiscussed earlier.

4. Conclusions

Nanocrystalline p-type semiconducting transparent CuAlO2thin film is synthesized on Si substrates via a DC sputteringtechnique. Proper phase formation of the material is confirmedby XRD measurements. HRTEM analysis shows the averagenanoparticle size to be around 25 nm with a film thicknessaround 100 nm. Electrical and Hall measurements reveala hole concentration of the order of 1018 cm−3 with aroom-temperature conductivity around 0.2 S cm−1. EDXanalysis indicates the presence of excess (non-stoichiometric)

6

Nanotechnology 24 (2013) 165705 A N Banerjee and S W Joo

oxygen in the material, which is considered to be the cause ofthe p-type nature of the nanomaterial. I–V characteristics ofthe nanostructured thin film reveals nonlinear behavior, whichis described in terms of field-assisted thermionic emissionfrom trap-states and explained through the Poole–Frenkelmodel. Adsorbed oxygen ions present in the grain boundaryregions act as the defect states and trap charge carriers toinhibit the electrical conductivity. Under the applied field andwith an increase in the sample temperature, the trap barrierpotential is reduced and thermal emission of charge carriersfrom the trap-states takes place to induce the nonlinearityin the I–V curves. The nonlinearity is found to dominateabove 50 V under our experimental conditions. Also, atransition from electrode-limited conduction at lower voltagesto sample-limited conduction at higher voltages is observedin the I–V curves, which is in accordance with the previousreports on oxide semiconductors. A relative comparison of thetrap energy levels with the activation energy and Fermi energyof our nanocrystalline CuAlO2 thin film shows the existenceof non-degenerate, thermally activated, semiconductor-typeconduction and also reveals the presence of deep trap levelsand a shallow acceptor level with an acceptor concentrationconsiderable higher than the trap-center concentration. Thisshows that a majority of the non-stoichiometric oxygen takespart in the hole generation, and only a small fraction acts asthe defect states near grain boundaries to trap charge carriersand hinder the hole conductivity.

Acknowledgments

This work was supported by World Class Univer-sity Grant No. R32-2008-000-20082-0 of the NationalResearch Foundation of Korea. Also we thank DrK K Chattopadhyay, Jadavpur University, India, for importantdiscussions.

References

[1] Kawazoe H, Yasukawa M, Yanagai H and Hosono H 1997Nature 389 939

[2] Banerjee A N and Chattopadhyay K K 2005 Prog. Cryst.Growth Charact. Mater. 50 52

[3] Tonooka K, Bando H and Aiura Y 2003 Thin Solid Films445 327

[4] Dittrich T, Dloczik L, Guminskaya T, Lux-Steiner M C,Grigorieva N and Urban I 2004 Appl. Phys. Lett. 85 742

[5] Banerjee A N, Nandy S, Ghosh C K andChattopadhyay K K 2007 Thin Solid Films 515 7324

[6] Banerjee A N and Chattopadhyay K K 2008 Cent. Eur. J.Phys. 6 57

[7] Yu R S and Yin H H 2012 Thin Solid Films 526 103–8[8] Banerjee A N, Joo S W and Min B K 2012 J. Appl. Phys.

112 114329[9] Banerjee A N and Chattopadhyay K K 2005 J. Appl. Phys.

97 084308[10] Laskowski R, Christensen N E, Blaha P and Palanivel B 2009

Phys. Rev. B 79 165209[11] Banerjee A N, Maity R, Ghosh P K and

Chattopadhyay K K 2005 Thin Solid Films 474 261[12] Kunihito K, Hisashi K and Seo W S 2001 J. Mater. Chem.

11 251[13] Liu C and Morelli D T 2011 J. Electron. Mater. 40 678

[14] Zhang H Y, Li P G, Chen C P, Tu Q Y and Tang W H 2005J. Alloys Compounds 396 40

[15] Mondelli C, Amrute A P, Schmidt T and Perez-Ramırez J2011 Chem. Commun. 47 7173

[16] Smith J R, Van Steenkiste T H and Wang X G 2009 Phys. Rev.B 79 041403

[17] Bandara J and Yasomanee J P 2007 Semicond. Sci. Technol.22 20

[18] Ding J et al 2010 Appl. Surf. Sci. 256 6441[19] Ding J, Sui Y, Fu W, Yang H, Zhao B and Li M 2011

Nanotechnology 22 295706[20] Banerjee A N and Joo S W 2011 Nanotechnology 22 365705[21] Banerjee A N and Chattopadhyay K K 2004 Appl. Surf. Sci.

225 243[22] Zheng X G, Taniguchi K, Takahashi A, Liu Y and

Xu C N 2004 Appl. Phys. Lett. 85 1728[23] Deng Z, Fang X, Li D, Zhou S, Tao R, Dong W, Wang T,

Meng G and Zhu X 2009 J. Alloys Compounds 484 619[24] Nakanishi A and Katayama-Yoshida H 2012 Solid State

Commun. 152 2078[25] Li J, Wang X, Shi S, Song X, Lv J, Cui J and Sun Z 2012

J. Am. Ceram. Soc. 95 431[26] Thomas G 1997 Nature 389 907[27] Banerjee A N, Ghosh C K and Chattopadhyay K K 2005 Sol.

Energy Mater. Sol. Cells 89 75[28] Yanagi H, Inoue S, Ueda K, Kawazoe H, Hosono H and

Hamada N 2000 J. Appl. Phys. 88 4159[29] Ingram B J, Mason T O, Asahi R, Park K T and

Freeman A J 2001 Phys. Rev. B 64 155114[30] Lee M S, Kim T Y and Kim D 2001 Appl. Phys. Lett. 79 2028[31] Banerjee A N, Kundoo S and Chattopadhyay K K 2003 Thin

Solid Films 440 5[32] Banerjee A N, Maity R and Chattopadhyay K K 2003 Mater.

Lett. 58 10[33] Banerjee A N, Ghosh C K, Das S and

Chattopadhyay K K 2005 Physica B 370 264[34] Pelto J, Paajanen M, Kannus K, Lahti K and Harju P 2004 Int.

Conf. on Solid Dielectrics (Toulouse, July)[35] 1999 Joint Committee on Powder Diffraction Standards—The

International Center for Diffraction Data (PA, USA:JCPDS-ICDD) File Card # 77-2493

[36] Banerjee A N, Chattopadhyay K K and Joo S W 2011 PhysicaB 406 220

[37] Cooper E A, Kirkpatrick K S, Christensen B J, Hong B S andMason T O 1992 Grain Boundary Controlled Properties ofFine Ceramics ed K Ishizaki, K Nihara, M Iostani andR G Ford (New York: Elsevier) pp 3–12

[38] Scalvi L V A, Messias F R, Souza A E, Li M S,Santilli C V and Pulcinelli S H 1998 J. Sol–Gel Sci.Technol. 13 793

[39] Banerjee A N, Maity R, Kundoo S andChattopadhyay K K 2004 Phys. Status Solidi a 201 983

[40] Ghosh P K, Maity R and Chattopadhyay K K 2004 Sol.Energy Mater. Sol. Cells 81 279

[41] Ghosh C K, Popuri S R, Sarkar D andChattopadhyay K K 2011 J. Mater. Sci. 46 1613

[42] Frenkel J 1938 Phys. Rev. 54 647[43] Chopra K L 1969 Thin Film Phenomenon (New York:

McGraw-Hill) p 506[44] Simmons J G 1970 Handbook of Thin Film Technology

(New York: McGraw-Hill) pp 14–31[45] Stuart M 1967 Phys. Status Solidi 23 595[46] Goswami A and Radhakrisnan S 1975 Indian J. Pure Appl.

Phys. 13 439[47] Simmons J G 1967 Phys. Rev. 155 657[48] Willardson R K and Beer A C 1971 Semiconductors and

Semimetals (New York: Academic)[49] Benko F A and Koffyberg F P 1984 J. Phys. Chem. Solids

45 57[50] Ashcroft N W and Mermin N D 2010 Solid State Physics

(New Delhi: Cengage Learning)

7