Four-point probe electrical resistivity scanning system for large area conductivity andactivation energy mappingKlimentiy Shimanovich, Yaniv Bouhadana, David A. Keller, Sven Rühle, Assaf Y. Anderson, and Arie Zaban

Citation: Review of Scientific Instruments 85, 055103 (2014); doi: 10.1063/1.4873353 View online: http://dx.doi.org/10.1063/1.4873353 View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/85/5?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Conduction mechanism of resistance switching in fully transparent MgO-based memory devices J. Appl. Phys. 114, 134301 (2013); 10.1063/1.4821900 Fluorine doped tin oxide (FTO) thin film as transparent conductive oxide (TCO) for photovoltaic applications AIP Conf. Proc. 1512, 710 (2013); 10.1063/1.4791235 The role of secondary phase precipitation on grain boundary electrical activity in Cu2ZnSnS4 (CZTS)photovoltaic absorber layer material J. Appl. Phys. 112, 124508 (2012); 10.1063/1.4769738 Microwave shielding of fluorine-doped tin oxide film obtained by spray pyrolysis studied by electricalcharacterization J. Appl. Phys. 105, 074502 (2009); 10.1063/1.3093690 Band gap tuning and high electrical conductivity in amorphous and polycrystalline films of the Cu x ( Cd Te ) y Oz system J. Appl. Phys. 100, 113713 (2006); 10.1063/1.2392718

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32

REVIEW OF SCIENTIFIC INSTRUMENTS 85, 055103 (2014)

Four-point probe electrical resistivity scanning system for large areaconductivity and activation energy mapping

Klimentiy Shimanovich, Yaniv Bouhadana, David A. Keller, Sven Rühle, Assaf Y. Anderson,and Arie Zabana)

Department of Chemistry, Bar Ilan University, Ramat Gan 52900, Israel

(Received 5 March 2014; accepted 14 April 2014; published online 5 May 2014)

The electrical properties of metal oxides play a crucial role in the development of new photovoltaic(PV) systems. Here we demonstrate a general approach for the determination and analysis of theseproperties in thin films of new metal oxide based PV materials. A high throughput electrical scanningsystem, which facilitates temperature dependent measurements at different atmospheres for highlyresistive samples, was designed and constructed. The instrument is capable of determining conduc-tivity and activation energy values for relatively large sample areas, of about 72 × 72 mm2, with theimplementation of geometrical correction factors. The efficiency of our scanning system was testedusing two different samples of CuO and commercially available Fluorine doped tin oxide coatedglass substrates. Our high throughput tool was able to identify the electrical properties of both resis-tive metal oxide thin film samples with high precision and accuracy. The scanning system enabledus to gain insight into transport mechanisms with novel compositions and to use those insights tomake smart choices when choosing materials for our multilayer thin film all oxide photovoltaic cells.© 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4873353]

I. INTRODUCTION

A. The need for high throughput tools

Over the past 30 years many useful discoveries in thefield of material science have been made as a result of adopt-ing high throughput approaches, particularly in the areas ofoptics,1–3 dielectrics,4 and magnetic materials.5, 6 This ap-proach offers an alternative to one by one or single step exper-iments by allowing multiple experiments to be performed in arelatively short time. Such techniques are generally achievedthrough the use of serial automation or by conducting severalmeasurements simultaneously.

Implementation of the high throughput approach in thefield of photovoltaics (PV) can significantly improve solarcell performance by optimizing properties of known mate-rials and by enabling the discovery of new semiconductingmaterials,7, 8 leading to new solar cell designs. Moreover thepresent PV market is growing rapidly and requires the devel-opment of new technological instruments for the fast charac-terization and analysis of solar cells and their components.

B. All oxide photovoltaics

All oxide PV,9–11 the new field in photovoltaic’s has gen-erated much interest due to the unique properties (chemicalstability, nontoxicity, and abundance) of many metal oxidesthat can allow for manufacturing under ambient conditionsand make them prime candidates for use as absorbers, electronhole transport layers, and transparent conducting oxides.12–15

To accelerate the discovery of viable new metal oxide mate-rials, we need a fast method of screening metal oxide proper-

a)Author to whom correspondence should be addressed. Electronic mail:arie.zaban@biu.ac.il

ties, suitable software for experimental control, data analysis,and experimental design and preparation of an array of mate-rials’ libraries to prepare, analyze, and interpret large numbersof diverse material compositions.3, 16–20 This can be achievedusing a high throughput approach. For this approach to be ef-fective there is a need for the large-scale material synthesis,optimization, and analysis for both new and known metal ox-ide semiconducting materials that are used for the generationand transportation of free charge carriers.

C. Description of high throughput system used for PVapplications

To perform high throughput screening and analysis ofmetal oxide materials for PV energy production21–24 we havedesigned and constructed a four point probe electrical scannerto analyze highly resistive samples of glass substrates coatedwith thin film metal oxides. The basic design incorporates:(1) a serial scanning instrument for resistance measurementwith a heated sample platform to produce homogeneousheating, (2) the ability to measure high resistance values ofup to 7 G� as a function of temperature up to 300 ◦C, and (3)uniquely implemented analytical calculations and softwaretools. The constructed system is capable of detecting regionsof interest in thin metal oxide film sample libraries made ofsingle, binary, ternary, or quaternary material compounds.The term “libraries” refers to a thin film sample, that can becreated with one or more variations of: thickness, chemicalcomposition, temperature, humidity, etc.

The effectiveness of our high throughput tool was testedon commercial Fluorine doped Tin oxide (FTO)28, 29 coatedglass substrate. We demonstrate electrical measurementswhich were performed on CuO thin film,25–27 on glass withthickness gradient. The samples were electrically analyzed

0034-6748/2014/85(5)/055103/6/$30.00 © 2014 AIP Publishing LLC85, 055103-1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32

055103-2 Shimanovich et al. Rev. Sci. Instrum. 85, 055103 (2014)

using resistivity,30 conductivity,31 and activation energy32

mapping. The correction factors33, 34 applied to activationand resistivity values obtained from our high throughputanalytical device made the characterization of the materials’electrical properties highly accurate. Thus our system joinsa set of tools for high throughput characterization of new PVmaterials.

II. EXPERIMENTAL SECTION

A. Fabrication of a high throughput electrical scanner

A temperature-controlled four point probe electrical re-sistivity scanning system was engineered by using a Z-arm(Olympus) to lift set of four probes up and down, an x-yscanning table with 100 mm x-range and 100 mm y-range(Märzhäuser Wetzlar GmbH & Co. KG) onto which a heatingstage was mounted to accommodate samples of up to 72 × 72mm2 and connected to a temperature controller (Eurotherm,model 3216). Then, a home-built linear four point probe headwith a constant inter probe spacing of 2.5 mm, achieved us-ing gold plated spring loaded probes (Ingun PrüfmittelbauGmbH), Keithley’s voltmeter and current supply, a hermeticPerspex box that is 70 cm in width, length, and height for hu-midity control, and an air temperature and humidity data log-ger (MRC, model 8808) were incorporated into the system.The system was controlled using a program that was designedin our lab using LabVIEW software. The scanner was testedon FTO and CuO coated glass substrates.

B. Preparation of FTO and CuO thin film coatings

Fluorine doped tin oxide (Transport Electron Conductor-TEC 15, Hartford) with a thickness of 2.2 mm and size of71.2 × 71.2 mm2 was used as a commercial reference sam-ple with known parameters (sheet resistance 15 �/square), forhigh throughput scanner validation. The homogeneous thick-ness profile of 440 nm of FTO coating was determined us-ing a Focused Ion Beam (FIB-FEI, model Helios 600) sys-tem. In order to remove impurities, the coated substrates wererinsed, sonicated, and cleaned and then used for electricalmeasurements.

CuO with an inhomogeneous thickness profile was pro-duced by pulsed laser deposition (PLD, Neocera) on to glasssubstrates supplied by Hartford (substrate size of 71.2 × 71.2mm2 and thickness of 2.2 mm). PLD CuO coating was pro-duced using a KrF laser with 8 Hz pulse rate, 97 mJ/pulseapplied energy, and ∼1.54 J/cm2 energy density. The mea-sured laser active area on the target was found to be 0.063 cm2

where the number of applied pulses was 30 000. The substratewas held in a fixed position in order to create CuO thin filmswith gradient thicknesses. The target substrate distance wasfixed at 77 mm with a constant oxygen pressure of 128 mTorrand temperature of 400 ◦C.

C. FTO and CuO thin film coatings using conventionalanalysis

In order to certify the accuracy of the results obtainedfrom our high throughput tool, coated samples were an-

alyzed and characterized using Energy Dispersive X-ray(EDX) and X-ray diffraction (XRD). The EDX analytical pa-rameters used for sample characterization were kept iden-tical to the high throughput tool parameters. The numberof CuO thickness data points remained 169, while sevencentral points were measured using a FIB for thicknessdata correction. The structural and electrical characteriza-tion of the films, deposited at optimized preparative param-eters, was carried out using X-ray diffraction that was per-formed on a Rigaku smart lab work station. The θ -2θ scanrange was performed from 30◦ to 70◦ for seven points onthe gradient thin film, measured at the central column of thesample.

III. RESULTS AND DISCUSSIONS

Characterization of metal oxides’ electrical propertiesprovides a better understanding of conduction mechanisms,which results in the efficient use of semiconductors in so-lar cells. This is why we chose to focus our research on theconstruction of new high throughput scanning system, whichenables rapid screening of materials’ properties and high pre-cision measurements in metal oxide sample characterization.The construction of a high throughput system started with se-lecting four point probes, which match system requirementssuch as appropriate scanning area, and high sensitivity, to en-able analysis of highly resistive samples. We equipped thesystem with two outer probes connected to a current supplier(Keithley Model 2400) and two inner probes connected to avoltmeter (Keithley 6517A Electrometer/High resistance me-ter, Figure 1) which enabled us to detect high thin film resis-tances up to ∼7 G�.

The heating stage was designed to enable homogeneousheat distribution throughout the sample interface by usingfour equally dispersed heaters positioned in parallel, in-side the sample holder, monitored with a thermocouple (seeFig. 1). The homogeneity of the heat distribution among thesample holder was detected using an IR camera at two sample

FIG. 1. Illustration of scanner system which consists of four point probehead, sample holder, four equally dispersed heaters with a thermocouple inthe middle, heat insulator plate, X-Y probing stage, and Z-arm. A picture ofthe high throughput electrical scanner is shown on the top left corner of theimage.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32

055103-3 Shimanovich et al. Rev. Sci. Instrum. 85, 055103 (2014)

FIG. 2. (a) Process of sample scanning, where the probes are lifted up, the X-Y scanning table moves to the next position and the probes come down to samplesurface again. (b) An illustration of probes position at the edges of the sample. (c) Correction factor map calculated using Eq. (3). (d) FTO resistivity as measuredat 47 ◦C. (e) Corrected resistivity of FTO sample at 47 ◦C; a product of (c) and (d).

holder temperature values of 50 ◦C and 200 ◦C, which weremeasured by thermocouple located inside the sample holderplate. The temperature readings on sample holder area showsdeviation of 0.6 ◦C throughout the holder, when sample holderwas heated to 200 ◦C.

A rectangular array of 13 rows and 13 columns, account-ing for a total of 169 semiconductor thin film points, wasemployed for two-dimensional sample mapping for resistanceevaluation, which is shown in Fig. 2(a).

The resistance of thin films was evaluated using thefour point probe technique, implemented in the scanner. Thismethod practically eliminates measurement errors due to con-tact resistances between the probes and the sample, whichusually exist in two point probe measurement technique.Moreover this technique is widely used in the semiconductorindustry to monitor the production process35, 36 and provideinformation on the various process steps37, 38 and active carrierconcentration of doped semiconductor films.39 The resistancemeasurement results were then used for conductivity valuecalculations. Conductivity rather the resistance is an intrin-sic physical property, independent of the particular shape orsize of the sample and important for the interpretation of so-lar cell performance. The resistivity and conductivity of FTO

and CuO thin films were calculated using Eqs. (1) and (2):

ρ = π t

ln2∗ (V/I) ∗ F, s � t (1)

σ = 1/ρ, (2)

where σ is the conductivity and ρ is the resistivity whichis a function of the measured film thickness t, current flow-ing through the sample I, voltage detected across the sampleV, geometrical correction factor F, and integration constantπ /ln(2) which is obtained when probes are uniformly spacedand the space is larger than film thickness.

Geometrical correction factors were applied for conduc-tivity values observed for the sample.40 The geometrical cor-rection factor expressed in Eqs. (3)–(5) is based on a theo-retical model for rectangular samples. The model was foundto effectively correct the conductivity/resistivity edge values(see Fig. 2). The geometrical correction factor F is evaluatedby the superposition of two geometrical correction factors: (1)Correction factor F⊥ applied to a sample when the scanner’sprobes are positioned perpendicular to the Y axis, and (2)correction factor F|| applied to a sample when the scanner’s

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32

055103-4 Shimanovich et al. Rev. Sci. Instrum. 85, 055103 (2014)

FIG. 3. (a) A 169 points, 2D plot of the CuO thickness gradient in the measured film. (b) Conductivity of the CuO film at 47 ◦C, including geometrical factorcorrection. (c) θ /2θ XRD measurements of the CuO thin film. The graphs’ colors correspond to the measured points indicated in the upright insert.

probes are positioned parallel to the X axis.

F = F⊥ ∗ F||, (3)

F⊥ = 1

1 + 12ln2 ln

(L2S

+2)(

L2S

+1)

(L2S

+ 52

)(L2S

+ 12

), (4)

F|| = 1

1 + 12ln2 ln

(L1/S)2+1(L1/S)2+1/4

. (5)

L1 and L2 are the distances from one of the outer probes tothe sample’s X and Y edges, respectively, S represents theconstant distance between probes (see Fig. 2(b)).

However, the accuracy of the measured resistance val-ues (obtained from the conductivity scanner) is affected bythe correlation between the thickness of the thin film’s coat-ing and the spacing between the scanner probes. In the caseswhere this correlation was above 40%, the conductivity cal-culations are indicative of undesired bulk materials and addi-tional correction is needed. In our system, due to the relativelywide 2.5 mm distance between the probes, the additional cor-rection factor will always be less than 40% and result in a fac-tor value of 1. Figure 2(c) shows the map of geometrical cor-rection factor values for rectangular samples which decreasealong the X-axis, due to the inequality in edge distances whereL1 < L2 (see Figure 2(b)). The high correction factor valuesobtained from the central part of the sample are found to be

0.99–1, while the lowest values were detected at the edge ofthe sample and vary from 0.89 to 0.97.

To demonstrate the performance of our system on metaloxide thin film coatings we chose to use a CuO semiconduc-tor. First, the CuO film thickness was measured. Conductivityis calculated using resistance and thickness values, thereforeaccuracy in thickness measurements is of paramount impor-tance. Figure 3(a) shows the map of thickness gradient valuesfor high quality copper oxide coating measured by an EDXscanning system and confirmed by a cross section image takenusing a FIB. The conductivity of CuO like other semicon-ductors increases with temperature due to the extension of itsFermi function which increases the number of charge carriers.Understanding temperature dependent conductivity is criticalfor understanding device performance, as well as for extract-ing fundamental information on conduction mechanisms andactivation energy for carrier transport. The activation energywas evaluated using Eqs. (6) and (7) in which the slope of asemi log plot of ln(σ ) versus T−1 equals −Ea/k (Arrheniusbehavior):

σ = σ0exp (−Ea/kT) , (6)

ln (σ ) = ln (σ0) − Ea

kT, (7)

where σ is the conductivity of the film, Ea is the activationenergy, and σ 0 is the pre exponential constant.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32

055103-5 Shimanovich et al. Rev. Sci. Instrum. 85, 055103 (2014)

FIG. 4. (a) Arrhenius plots along gradient. (b), (c) Maps of activation energy calculated at temperatures range of 167 ◦C – 247 ◦C and 28 ◦C – 167 ◦C, respec-tively.

Figure 3(b) shows the map of corrected conductivityvalues which vary from 1×10−4 to 4.5×10−4 [S/cm]. Suchsmall variations in conductivity are consistent with the XRDdata (see Figure 3(c)), predominantly showing the forma-tion of CuO, which appears in cubic structures, for the en-tire sample and an increase in planes orientation with in-crease in film thickness. The conductivity values higher than4.5×10−4 [S/cm] are considered a measurement error due tothe deformation of the film at thicknesses lower than ∼30 nm,where the pressure has a major effect and deform the thinfilm.36

Figure 4(a) shows Arrhenius plots for temperatures vary-ing between 28 ◦C and 247 ◦C revealing two regimes of acti-vation energies that can correspond to two main conductionmechanisms: (1) Electron hopping from one trap to the otherat low temperature range, and (2) intrinsic conduction domi-nant at high temperature zone.41 Therefore, the activation en-ergy for the entire sample was calculated and is presented forthe two regions (Figures 4(b) and 4(c)). For temperatures be-tween 28 ◦C and 167 ◦C the activation energy remains identi-cal throughout the sample area with a value of 0.28 eV and anerror of kbT ∼= 0.026 eV. In the temperature range of 187 ◦Cto 247 ◦C the activation energy is not uniform, increasing tovalues between 0.45 eV and 0.65 eV.

IV. CONCLUSIONS

We demonstrated the development of a resistivity, con-ductivity, and activation energy mapping instrument. The ef-fectiveness of the instrument was probed on cuprous oxidethin films with inhomogeneous thicknesses and a commercial

FTO homogeneously coated sample. The mapping instrumentwas designed to detect high thin film resistances at the rangeof ∼7 G�. It contains geometrical correction factors therebyproviding high accuracy including the sample edge. The sys-tem can operate under different atmospheric environments ina temperature range of 27 ◦C to 300 ◦C. The designed instru-ment provides means for determining the mechanisms of elec-trical conduction in the sample mainly in high throughputstudies.

ACKNOWLEDGMENTS

This study was funded by the European Commission un-der the FP7 All Oxide PV project entitled “Novel CompositeOxides by Combinatorial Material Synthesis for Next Gener-ation All-Oxide-Photovoltaics” number 309018.

1Z. Zou, Y. Liu, H. Li, Y. Liao, and C. Xie, Comb. Chem. 12, 363 (2010).2Q. Wang, J. Perkins, M. H. Branz, J. Aleman, C. Duncan, and D. Ginley,Appl. Surf. Sci. 189, 271 (2002).

3J. D. Perkins, C. W. Teplin, M. F. A. M. Van Hest, J. L. Alleman, X. Li, M.S. Dabney, B. M. Keyes, L. M. Gedvilas, D. S. Ginley, Y. Lin, and Y. Lu,Appl. Surf. Sci. 223, 124 (2004).

4H. Chang, C. Gao, I. Takeuchi, Y. Yoo, J. Wang, P. G. Schultz, and X. D.Xiang, Appl. Phys. Lett. 72, 2185 (1998).

5I. Takeuchi, R. B. van Dover, and H. Koinuma, MRS Bull. 27, 301 (2002).6Y. K. Yoo, Q. Xue, H. Lee, S. Cheng, X. D. Xiang, G. F. Dionne, S. Xu,J. He, Y. S. Chu, S. D. Preite, S. E. Lofland, and I. Takeuchi, Appl. Phys.Lett. 86, 042506 (2005).

7A. Zakutayev, J. D. Perkins, P. A. Parilla, N. E. Widjonarko, A. K. Sigdel,J. J. Berry, and D. S. Ginley, MRS Commun. 1, 23 (2011).

8A. Zakutayev, T. R. Paudel, P. F. Ndione, J. D. Perkins, S. Lany, A. Zaunger,and D. S. Ginley, Phys. Rev. B 85, 085204 (2012).

9S. Ruhle, A. Y. Anderson, H. N. Barad, B. Kupfer, Y. Bouhadana, E. Rosh-Hodesh, and A. Zaban, Phys. Chem. Lett. 3, 3755 (2012).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32

055103-6 Shimanovich et al. Rev. Sci. Instrum. 85, 055103 (2014)

10A. Y. Anderson, Y. Bouhadana, H. N. Barad, B. Kupfer, E. Rosh-Hodesh,H. Aviv, Y. R. Tischler, S. Rühle, and A. Zaban, ACS Comb. Sci. 16, 53(2014).

11B. D. Yuhas and P. Yang, J. Am. Chem. Soc. 131, 3756 (2009).12A. N. Banerjee and K. K. Chattopadhyay, Prog. Cryst. Growth Charact.

Mater. 50, 52 (2005).13C. M. McShane and K. S. Choi, Phys. Chem. Chem. Phys. 14, 6112

(2012).14H. Kidowaki, T. Oku, and T. Akiyama, Int. J. Res. Rev. Appl. Sci. 13, 67

(2012).15C. G. Granqvist, Sol. Energy Mater. Sol. Cells 91, 1529 (2007).16H. M. Christen, S. D. Silliman, and K. S. Harshavardhan, Rev. Sci. Instrum.

72, 2673 (2001).17R. E. Treharne, K. Hutchings, D. A. Lamb, S. J. C. Irvine, D. Lane, and K.

Durose, J. Phys. D: Appl. Phys. 45, 335102 (2012).18S. Borhani-Haghighi, M. Kieschnic, Y. Motemani, A. Savan, D. Rogalla,

H. W. Becker, J. Meijer, and A. Ludwing, ACS Comb. Sci. 15, 401(2013).

19O. J. Gregory, I. M. Tougas, M. Amani, and E. E. Crisman, ACS Comb.Sci. 15, 580 (2013).

20A. Kafizas and I. P. Parkin, Chem. Soc. Rev. 41, 738 (2012).21A. Zakutayev, J. D. Perkins, P. A. Parilla, N. E. Widjonarko, A. K. Sigdel,

J. J. Berry, and D. S. Ginley, Rev. Sci. Instrum. 84, 053905 (2013).22T. Minami, Y. Nishi, and T. Miyata, Appl. Phys. Express 6, 044101 (2013).23H. M. Wei, H. B. Gong, L. Chen, M. Zi, and B. Q. Cao, J. Phys. Chem. C

116, 10510 (2012).

24K. C. Hewitt, P. A. Casey, R. J. Sanderson, M. A. White, and R. Sun, Rev.Sci. Instrum. 76, 093906 (2005).

25A. Y. Oral, E. Mensur, M. H. Aslan, and E. Basaran, Mater. Chem. Phys.83, 140 (2004).

26A. Chen, G. Yand, H. Long, F. Li, Y. Li, and P. Lu, Thin Solid Films 517,4277 (2009).

27N. Serin, T. Serin, S. Horzum, and Y. Çelik, Semicond. Sci. Technol. 20,398 (2005).

28A. E. Rakhshani, Y. Makdisi, and H. A. Ramazaniyan, J. Appl. Phys. 83,1049 (1998).

29S. Shanthi, C. Subramanian, and P. Ramasamy, Cryst. Res. Technol. 34,1037 (1999).

30A. Mircea, Sci. Instrum. 679, 3 (1964).31R. K. Franks and J. B. Robertson, NASA Tech. Note TN D-6504 (1971).32Y. Mizuno, M. Okubo, E. Hosono, T. Kudo, H. Zhou, and K. Oh-ishi, J.

Phys. Chem. C 117, 10877 (2013).33A. Mircea, Solid-State Electron. 6, 459 (1963).34L. B. Valdes, Proc. IRE 40, 1429 (1952).35L. J. Swartzendruber, Solid-State Electron. 7, 413 (1964).36P. J. Severin, Solid-State Electron. 14, 247 (1971).37S. C. Choo and M. S. Leong, Solid-State Electron. 22, 405 (1979).38H. H. Berger, J. Electrochem. Soc. 119, 507 (1972).39D. S. Perloff, F. E. Wahl, and J. Conragan, J. Electrochem. Soc. 124, 582

(1977).40F. M. Smits, Bell Syst. Tech. J. 37, 711 (1958).41Y. K. Jeong and G. M. Choi, J. Phys. Chem. Solids 57, 81 (1996).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

87.68.230.0 On: Mon, 05 May 2014 15:33:32