Electrical properties of thin films formed on self ...

8 downloads 0 Views 395KB Size Report
This can be understood by the Poole-. Frenkel conduction mechanism, as described in Sec. III D. Figure 5 illustrates a schematic energy band diagram of.
Electrical properties of on silicon

thin films formed on self-assembled organic monolayers

Hyunjung ShinMark R. De Guire and Arthur H. Heuer

Citation: Journal of Applied Physics 83, 3311 (1998); doi: 10.1063/1.367132 View online: http://dx.doi.org/10.1063/1.367132 View Table of Contents: http://aip.scitation.org/toc/jap/83/6 Published by the American Institute of Physics

JOURNAL OF APPLIED PHYSICS

VOLUME 83, NUMBER 6

15 MARCH 1998

Electrical properties of TiO2 thin films formed on self-assembled organic monolayers on silicon Hyunjung Shin Samsung Advanced Institute of Technology, P.O. Box 111, Suwon 440-600, Korea, and Department of Materials Science and Engineering, Case Western Reserve University, Cleveland, Ohio 44106

Mark R. De Guire and Arthur H. Heuer Department of Materials Science and Engineering, Case Western Reserve University, Cleveland, Ohio 44106

~Received 4 August 1997; accepted for publication 11 December 1997! TiO2 thin films were synthesized onto sulfonated self-assembled organic monolayers from Ti41 aqueous solutions at low temperature ~80 °C!. The deposited TiO2 thin films were characterized using Rutherford backscattering spectroscopy and transmission electron microscopy. The electrical properties of the TiO2 films were measured using capacitance–voltage (C – V) and current–voltage techniques. From C – V measurements, the dielectric constants were calculated to be in the range of 24–57 for the as-deposited films and 67–97 for films annealed at 400 °C. For the as-deposited samples, the breakdown voltage was 2.7 MV/cm, the leakage current was 4.531026 A/cm2, the resistivity was 1 – 1.53109 V cm, and the interface trap density ;131012 cm22 eV21. The leakage current behavior was consistent with the Poole-Frenkel mechanism of conduction. © 1998 American Institute of Physics. @S0021-8979~98!08206-1#

I. INTRODUCTION

High resistivity, high breakdown voltage, and low leakage current, which are closely related to carrier trapping and conduction processes, are required properties for capacitor applications. These properties are also affected by the defects within, composition of, and microstructure of the oxide films. Besides the dielectric properties, which are of primary concern for storage capacitors, fixed oxide charge and interface trap density also require thorough investigation in any dielectric being considered for the gate oxide in transistors. Fixed charge is a property of the dielectric material, and its density is independent of band bending and surface potential. Unlike fixed charges, the interface traps exchange charge with the silicon substrate when the surface potential is varied. They can be charged and discharged by varying the gate voltage. In the present work, a novel low-temperature processing technique—in essence, a ‘‘biomimetic’’ approach13–16—was used to synthesize TiO2 thin films. Using Ti~IV! aqueous solutions, TiO2 films with the anatase structure were formed on functionalized self-assembled organic monolayers17 at low temperatures ~80 °C!. The present process may have significant advantages over conventional vacuum deposition techniques for the formation of oxide thin films. Vapor techniques primarily involve line-of-sight deposition, whereas our solution process should be conformal and applicable to complex shapes and a wide variety of surfaces. Most importantly, achievement of the desired properties with vapor deposition routes often requires that the substrate be heated ~often to several hundreds of °C!, either during deposition or subsequently, to convert the usually amorphous as-deposited material into a well-ordered crystalline film. The thermal demands of the present technique are, in contrast, quite modest.

The development of future ultralarge scale integration ~ULSI! storage capacitors for dynamic random access memory ~DRAM! requires new dielectric materials. The capacitance, C, of a dielectric layer is given by e 0 e r A/d where e 0 is the permittivity of vacuum, e r is the relative permittivity of the dielectric layer, A is the electrode area, and d is the thickness of the dielectric layer. Increasing the capacitance per unit area of dielectric requires decreasing d or increasing e r . Conventional SiO2 layers, with e r 53.9, are limited to a minimum thickness of about 5 nm because the tunneling current and pinhole density increases dramatically for smaller thicknesses. To overcome this problem, a new highe r material showing sufficiently low leakage current densities is necessary for improved memory operation.1 Large capacitance can be desirable even for gate dielectrics in transistors. For a fixed gate oxide thickness, d, the applied gate voltage can be smaller, and still give a charge density at the gate-channel interface sufficient for operation of the device. TiO2 is a very promising insulator for application to memory devices, as it exhibits a higher dielectric constant ~up to about 100! than Ta2O5 with the same breakdown field strengths.2 It is also physically and chemically stable, and several investigators produced TiO2 thin films using various techniques and measured their electrical properties.3–12 It should be noted that interfacial reactions between semiconductors and oxide films at high temperatures during processing may adversely affect the electrical properties of dielectric films, because the reacted oxide layer may have undesirable charge traps. Chemical preparation at low temperatures,3,4 which is utilized in this work, reduces the possibilities of diffusion and intermixing at the semiconductor-oxide interface. 0021-8979/98/83(6)/3311/7/$15.00

3311

© 1998 American Institute of Physics

3312

J. Appl. Phys., Vol. 83, No. 6, 15 March 1998

Shin, De Guire, and Heuer

In this article, we demonstrate the potential application of the deposited TiO2 films as gate insulators in semiconductors. Capacitance–voltage (C – V) and current–voltage (I – V) measurements were used to evaluate TiO2 thin films deposited on self-assembled monolayers ~SAMs!. The structures which have been generated—a metal electrode evaporated onto the TiO2 film on ~the SAM layer! silicon— comprise a metal-oxide-semiconductor ~MOS! capacitor. The expected behavior of such structures is well established, giving a basis from which the interfacial properties of the samples can be inferred. Dielectric constants, resistivity, breakdown voltages, and leakage currents were measured and are reported here. A mechanism for electrical conduction in these TiO2 films on SAMs is proposed. II. EXPERIMENTAL PROCEDURE A. Synthesis of TiO2 films

The substrates used in this article were p-type $001% single crystal Si wafers. These wafers were cleaned and oxidized ~to form 1.5 nm of hydrated SiO2! in ‘‘piranha’’ solution ~75 vol % H2SO4, 7.5 vol % H2O2, and 17.5 vol % H2O!, and dried in air. They were then dipped into dicyclohexyl ~an organic solvent! containing 1 vol % of the active surfactant, trichlorosilylhexadecane thioacetate, during which the SAM formed spontaneously. After thorough washing to remove all traces of the organic solvent, the thioacetate functionality of the SAMs was oxidized to a sulfonate (SO3H) functionality by immersing the functionalized Si wafers in oxone, an aqueous persulfate oxidant. ~Further details of the synthesis of sulfonated SAMs on Si can be found elsewhere.18! To form the TiO2 thin films,19,20 reagent grade titanium tetrachloride (TiCl4) was used as starting material and was purified by distillation at 68–71 °C, the exact temperature depending on house vacuum. The purified TiCl4 was colorless and was added slowly (;1 ml min21) with stirring into a 6 M HCl aqueous solution that was cooled in an ice bath (0 °C) to prepare a 0.5 M TiCl4 aqueous solution which was then filtered. A portion ~25 ml! of this stock solution was charged into a 150 ml beaker and covered with a glass dish. All glassware had been dried in a drying oven to minimize the occurrence of unintended hydrolysis. The silicon wafers coated with sulfonated SAMs were immersed into the TiCl4 solution at 80 °C. The temperature was maintained constant (61 °C) by a Thermo-watch regulator with an oil bath. Typical film deposition times were 0.5–4 h. After removal from the solutions, the TiO2-coated samples were washed with distilled water and dried in air. Some of the films were heated in air at 10 °C/min to 400 °C and held for 2 h, then cooled in the furnace. B. Structural characterization and thickness measurements of the TiO2 films

Rutherford backscattering spectroscopy ~RBS! using a NEC 5SDH ion beam accelerator and analytical transmission electron microscopy ~TEM! using a Philips CM-20 equipped with a high-purity ~intrinsic! Ge Noran energy dispersive spectroscopy ~EDS! detector were performed on the films. Thin foils for TEM analysis were made by cutting the Si

FIG. 1. ~a! Cross-sectional TEM micrograph showing a film thickness of 20.563 nm. ~b! RBS spectrum from same sample, yielding film thickness of 19 nm. ~For thickness determination using RBS, the films were assumed to be pure and stoichiometric TiO2 with the density of bulk anatase, 3.9 g/cm3.!

substrates bearing the TiO2 thin films in half and gluing the TiO2-coated faces together using epoxy resin. The thickness of the specimen was reduced to less than 25 mm by hand polishing. The sample was then mounted on a 3 mm copper grid and further thinned using conventional ion thinning. The films formed on sulfonated SAMs contained fine-grained anatase ~2–3 nm in diameter! and a small quantity (,25%) of amorphous material;19 they were uniform in thickness, pore free, and well packed. The interfaces between films and substrates were well defined, flat, and planar. Film thicknesses were also measured using ellipsometry, which data were checked using RBS and cross-sectional TEM techniques. For example, the cross-sectioned sample in Fig. 1~a! shows a film thickness of 20.563 nm, while the thickness of this specimen obtained from the RBS spectrum, Fig. 1~b!, was 19 nm. ~For thickness determination using RBS, the films were assumed to be pure and stoichiometric TiO2 with the density of bulk anatase, 3.9 g/cm3. The small amount of chlorine present ~less than 1 w/o! was ignored in this analysis. We note, however, that chlorine incorporation can improve the dielectric properties of the films by sodium passivation.21!

J. Appl. Phys., Vol. 83, No. 6, 15 March 1998

Shin, De Guire, and Heuer

3313

C. C – V and I – V measurements

The relative dielectric constant, breakdown voltage, and leakage current of the TiO2 thin films were obtained from C – V and I – V measurements. For this purpose, aluminum electrodes were evaporated onto the TiO2 films and on the back of the silicon wafers, resulting in a MOS capacitor structure. The devices are assumed to have ohmic contacts between the semiconductor and metal. The electrodes on the TiO2 films consisted of an array of circular aluminum spots, whose diameter was measured using optical microscopy. Electrodes of 0.5 mm2 were used for the as-deposited films, whereas smaller electrodes (0.1 mm2) were required for the annealed films to assure that the total capacitance was within the measurement limits of our C – V apparatus ~2 nF!. For C – V measurements, a sinusoidal ac voltage of various frequencies ~typically around 1 MHz22! was superimposed on a dc voltage that was varied between 1 and 25 V using a Keithley 590 high frequency C – V meter. All measurements were done at room temperature. The breakdown voltage and leakage current were measured with a Keithley I – V analyzer.

III. RESULTS AND DISCUSSION A. Dielectric constant

A typical C – V curve for one of the as-deposited Al/TiO2~SAM1SiO2!/p-Si samples is shown in Fig. 2~a!. Total capacitance in the accumulation region ~measured at more than ten individual electrode locations on the oxide layer! ranged from 1.1 to 1.8 nF. The total capacitance in the accumulation region, C Tot , is modeled as consisting of three capacitances in series i.e., those of the TiO2 layer, C TiO2, the SAM layer, C SAM , and the SiO2 layer, C SiO2 , to which the SAM is attached. Therefore, the total capacitance is given by 1 1 1 1 5 1 1 C Tot C SAM C SiO2 C TiO2

~1a!

which can be rearranged to give C TiO25

C Tot–C SiO2–C SAM C SiO2–C SAM2C Tot~ C SAM1C SiO2!

.

~1b!

The various capacitances are calculated from the equation C5

e 0 –e r –A , d

~2!

where e r and d are, respectively, the dielectric constant and the thickness of the SiO2 layer or the SAM layer, e 0 is the permittivity of free space, and A is the area of the electrode spot being tested. Assuming the values e r,SiO253.9,23 e r,SAM52.5,24 d SiO251.5 nm, d SAM52.5 nm, and taking d TiO2563 nm from ellipsometric measurements for this particular specimen, the relative dielectric constants of the asdeposited TiO2 layer were in the range 24–57. Figure 2~b! shows a C – V curve for a TiO2 film annealed at 400 °C for 2 h in air. The total capacitance ranged from 0.68 to 0.8031029 F at various locations on the oxide. More

FIG. 2. ~a! C – V curve for the as-deposited Al/TiO2~SAM1SiO2!/ p-Si sample ~b! C – V curve for a TiO2 film on p-Si, annealed at 400 °C for 2 h in air.

than ten spots were measured. Since the SAM layer is known to have been pyrolyzed at or below 400 °C,25 the capacitance of the annealed sample is determined by the TiO2 layer and the SiO2 layer in series. Taking d TiO2543 nm ~from ellipsometric measurements! and assuming d SiO252.5 nm ~some modest oxidation is assumed to have occurred during annealing26!, the dielectric constants for the annealed TiO2 layer ranged from 67 to 97, values comparable to the permittivity of bulk polycrystalline TiO2. The higher dielectric constant in the annealed film is assumed to result from some film densification and crystallization of the minor amorphous phase in the as-deposited films27 that occurred during the heat treatment. B. Interface and fixed-oxide trap charge density

Besides the charge induced by the applied field, there are two basic types of charges in the oxide layer: ~1! interface trap charge, Q it , and ~2! oxide charge, Q o . The interface trap charge varies with gate bias, whereas oxide charge is independent of gate bias. The oxide charge includes three types of charges: oxide fixed charge, Q f , oxide trap charge, Q ot , and mobile ionic charge, Q m . Each of these is discussed in more detail below.

3314

J. Appl. Phys., Vol. 83, No. 6, 15 March 1998

Shin, De Guire, and Heuer

~1! Interface trap charges (Q it) are due to the dangling bonds at the Si–SiO2 interface and are dependent on the chemical composition of this interface. The interface traps can be charged and discharged by varying the applied voltage. The charged and discharged interface trap centers at the interface will disrupt the distribution of the electric field, which is induced from charges in the semiconductor or the metal electrode, by terminating the electric lines across the oxide layer. ~2a! Oxide fixed charge (Q f ) is a property of the oxide layer and its density, and is independent of band bending and surface potential. This change can be due to localized charge centers that cannot change state by exchange of mobile carriers with the silicon. Also, this charge is located within approximately 3 nm of the Si-oxide interface and is due to unsatisfied silicon bonds in the oxide near the interface. Both fixed charge and interface trap charges are interface related and are sensitive to the device processing. In particular, the interface trap charges must be controlled and minimized ~to the low 1010 cm22 eV21 range! for reliable device performance. ~2b! For polycrystalline oxide films as the insulating layer in MOS devices, oxide-trapped charge (Q ot), which arises from various defects ~e.g., grain boundaries, open pores, etc.! in the films, may have the most significant influence on device performance. ~2c! The mobile ionic charge (Q m ) most commonly is caused by the presence of ionized alkali metal impurities such as sodium or potassium. This type of charge can drift in an applied field and result in device failure. Because the magnitude of Q o does not vary with applied voltage, and because it will either add to or subtract from any field applied to the interface, it will appear as an offset on the voltage axis of a C – V curve. This horizontal shift in a C – V curve is the simplest and most widely used method for measuring oxide charge density. If the difference in work function between the metal and the semiconductor, f ms , is not zero, and if the values of Q f , Q m , and Q ot are significant, the shift in the experimental capacitance–voltage curve, called the flatband voltage V FB , will be V FB5fms2

Q f 1Q m 1Q ot , C ox

~3!

where C ox is the capacitance (F/cm2) of the oxide, including that due to Q 0 . ~Note that this shift measures total oxide charge; it cannot distinguish between Q f , Q m , and Q ot .! The value of the flatband voltage V FB in the unannealed film was 20.64 V, suggesting the presence of positive charges in the oxide film. The positive charges could result from oxygen vacancies (V 0¨ ) due to dangling and/or broken bonds at the interface between the TiO2 crystallites and the amorphous TiO2 present in the films. From Eq. ~3!, the total oxide charge was in the range of 29.4 to 25.7 31028 C/cm2, taking fms520.9 V and V FB520.64 V. The sample annealed at 400 °C @Fig. 2~b!# showed a further shift of the C – V curve compared to the as-deposited sample in the direction of negative bias. The V FB increase to 21.2 V reflects an increase in the density of defect states in the insulator. This could have resulted from the formation of

FIG. 3. A C – V curve for the as-deposited Al/TiO2~SAM1SiO2!/ p-Si sample with theoretical C – V curve.

distinct grain boundaries, thus creating more dangling bonds. The formation of a new interface between the TiO2 and SiO2 layers after the decomposition of the SAM layer could also provide more interfacial oxide traps. These kinds of trapped charges have been studied in other double-layer dielectrics, for example, Al–Y2O3 –SiO2-Si, where it was proposed that negative trapped charges were formed at the Y2O3 /SiO2 interface.28 Interface trap density can be measured by a highfrequency method developed by Terman.29 The method relies on a C – V measurement at a frequency sufficiently high that interface traps are assumed not to respond to the ac probe but do respond to the varying dc voltage, and cause the C – V curve to stretch out along the voltage axis. The stretch out produces a nonparallel shift of the C – V curve. The interface trap density (D it) is determined from the difference between the ideal and experimental f s vs V curves by D it5

C ox d ~ DV G ! • , q dfS

~4!

where DV G is the voltage shift of the experimental curve with respect to ideal behavior at each value of C, C ox is the oxide capacitance, f s is the surface potential ~the total potential drop across the semiconductor, measured from the surface to the bulk reference!, and q is the charge on an electron. The f s vs V G curve can be constructed from highfrequency C – V curves. ~Details of such construction are given in Ref. 30.! Figure 3 shows both the experimental and ideal C – V curves for the as-deposited TiO2 thin film of Fig. 2~a!. An interface trap density of 131012 cm22 eV21 was calculated for the as-deposited sample. C. Leakage current and breakdown voltage

Figures 4~a! and 4~b! show the leakage current density versus applied electric field for the as-deposited ~a! and annealed ~b! TiO2 films in the accumulation mode forward bias ~i.e., with the Al contact biased negatively!. The leakage current was at least two orders of magnitude higher in the an-

J. Appl. Phys., Vol. 83, No. 6, 15 March 1998

Shin, De Guire, and Heuer

3315

FIG. 5. The energy band diagram of the Al/TiO2/SiO2 / p-Si MOS structure.

FIG. 4. Plots of leakage current density vs applied electric field for the as-deposited ~a! and annealed ~b! TiO2 films at the forward bias.

nealed sample at all applied fields. The resistivities, r, of the films were calculated from the I – V data @ I5V/R⇒ r 5 (V•A)/(I•d) # . The as-deposited films had a resistivity of 1.0– 1.53109 V cm, while those of the annealed film were ;108 V cm. Due to the microstructural changes, mainly coarsening of the grains in the TiO2 films, a ‘‘rougher’’ interface between the Al electrode and the TiO2 film surface can be expected to be present in the annealed sample. Locally, this roughness results in the generation of weak spots at the contact, which serve as high field regions. This enhances the injection of carriers and induces increased conduction. This phenomenon has also been observed in rough SiO2 films.31 Grain boundaries of polycrystalline TiO2 films probably become paths for leakage current, as occurs in other polycrystalline oxide layers.32 Also, a new interface between the TiO2 and SiO2 layers must be formed as the SAM layer is decomposed. The abrupt discontinuity at the interface could form other defect states in the band gap of the oxide. Such defect structures in the grain boundaries and the SiO2 /TiO2 interface could act as charge traps. Therefore, the leakage current is expected to be a bulk-limited process which is related to defects in the bulk. This can be understood by the PooleFrenkel conduction mechanism, as described in Sec. III D. Figure 5 illustrates a schematic energy band diagram of the Al/TiO2 /SiO2 / p-Si MOS structures when the Al contact is negatively biased. Electrons can be injected into defect states in the TiO2 insulator. Because these specimens had thin SiO2 layers (,2.5 nm), it is possible that some of the injected carriers could have sufficient energy to tunnel through to the silicon from the defect states. This could have

caused the observed current leakage and breakdown through the TiO2 films. The leakage current density obtained with the Al biased negative ~i.e., in the accumulation mode! was always lower ~for the same field! than the values measured with the Al biased positive. The leakage current of the as-deposited TiO2 films was 4.531026 A/cm2 at a field of 1 MV/cm. ~Similar behavior is observed in other oxide systems.33,34! This implies that for the case of Al/TiO2 /Si structures, differences in barrier height between the two different injecting carriers ~electrons when Al is biased negatively, and holes when Al is biased positively! do influence the leakage current which passes through the dielectric film. Therefore, the Al/TiO2 /p-Si structures showed rectifying characteristics. When the top Al electrode was biased positively with respect to the p-silicon ~i.e., the depletion and inversion conditions!, a different leakage current behavior was observed ~Fig. 6!. At reverse bias, breakdown occurred fairly abruptly, at 2.7 MV/cm. However, this dielectric breakdown behavior, with its quasisaturation regime at high field (.3 MV/cm),

FIG. 6. A plot of leakage current density ~J! vs applied electric field ~E! for the as-deposited TiO2 film at reverse bias.

3316

J. Appl. Phys., Vol. 83, No. 6, 15 March 1998

Shin, De Guire, and Heuer

appears to be different from the normal breakdown, which shows no high-field saturation. This suggests that the observed current–voltage behavior in the inversion mode is due to hole trapping in localized regions of the TiO2 film/SAM/SiO2 interface~s!. The large leakage current flowing through these TiO2 films is a significant problem, limiting their application as storage dielectrics. Other promising high-dielectric materials, for example, Y2O335,36 and Ta2O5, 37,38 have similar difficulties. However, the measured leakage current of as-deposited TiO2 films in this work is two orders of magnitude higher than that of Ta2O539,40 at the same electric field ~1 MV/cm!.

D. Conduction mechanisms in TiO2 films

It is known that grain boundaries in dielectric ceramics exhibit even higher specific resistivities than do the interior of the grains.41 After a dc voltage is applied, a large portion of the applied voltage appears across the grain boundaries. These local electric fields ~which can be as large as 105 V/cm!42 lead to local dielectric breakdown or fieldassisted emission of trapped charge carriers ~the PooleFrenkel effect!. Under these circumstances, electron or hole currents increase due to ionized donors or traps at grain surfaces, in the bulk, and also at the electrode. Importantly, the ionization energy is inversely proportional to the square of the permittivity. Thus, this effect in low k materials is far less significant than in medium- and high-e r materials.43 The Poole-Frenkel relation can be used to describe the forward current–voltage behavior of MOS structures J5CE exp$ 2q @ w 0 2 ~ qE/ p e effe 0 ! 1/2# /kT % 5CE exp~ 2q w 0 /kT ! exp~ b E 1/2! ,

~5a!

where E is the electrical field, C is a constant related to trap density, q is the charge on an electron, w 0 is the barrier height, and b 5q 3/2 / Ap e effe0kT. Rearranging Eq. ~5a! yields ln

SD

J qw0 1 b E 1/2 5ln C2 E k•T

~5b!

so that a plot of ln(J/E) vs E 1/2 , should be linear with slope b and an intercept equal to ln C2(qw0 /kT). Typical ln(J/E) 1 2

vs E plots for the as-deposited and 400 °C annealed TiO2 films are shown in Figs. 7~a! and 7~b!. A linear dependence is observed at high fields (.0.3 MV/cm). The deviation from the straight line at lower fields could be explained by the presence of positive oxide charges resulting in a naturally depleted surface at the substrate/oxide interface. The current increases exponentially with the square root of the field, implying a bulk-limited conduction process. From the linear portion of the plot, the calculated e eff values for as-deposited TiO2 film was 30, in the range of e r values determined from analysis of C – V and I – V data ~Sec. III A!. This agreement supports the conclusion that the dominant conduction mechanism through the TiO2 was bulk-limited rather than electrode-limited, and the linear behavior of ln(J/E) vs E 1/2 indicates that the bulk conduction is described by the Poole-

FIG. 7. The ln(J/E) vs E 1/2 plots for as-deposited ~a! and annealed at 400 °C ~b! TiO2 films.

Frenkel mechanism. Unlike amorphous SiO2 layers, other polycrystalline oxide layers such as Y2O336 and Ta2O544 show the same conduction process. It does not seem likely that oxygen vacancies (V 0¨ ) are the mobile species undergoing electromigration under a dc applied voltage. The mobility of oxygen vacancies at room temperature will be very small compared to that of electronic charge carriers. First, ionic transport has a large effective thermal activation barrier. Second, grain boundaries represent barriers of significant magnitude for the migration of V 0¨ . 45 Because of this, it is usually assumed that vacancies are immobile at low temperatures.46 IV. SUMMARY

The electrical properties of the as-deposited and annealed TiO2 thin films were characterized by C – V and I – V measurements ~Table I!. From these data, it was inferred that the annealed films contain a higher density of defect states.

TABLE I. Summary of the electrical properties of TiO2 thin films synthesized in this work.

Dielectric constant Resistivity ~V cm! Leakage current (A/cm2) Breakdown voltage ~MV/cm!

As-deposited

Annealed ~400 °C!

24–57 ;109 4.531026 2.7

67–97 ;108 531024 1.5

J. Appl. Phys., Vol. 83, No. 6, 15 March 1998

The current transport appeared to be dominated by bulklimited processes ~the Poole-Frenkel conduction mechanism!. Compared to Ta2O5 ( e r 58;17) 25,27,30–33,35,39,47,48 and Y2O3 ( e r 510;27) 28,29,34,49–51 as improved dielectrics relative to SiO2, the TiO2 films produced here had dielectric constants that were 2–3 times higher. TiO2 films could, therefore, provide higher charge storage capacity than present dielectrics. The variations of relative permittivity in the present TiO2 layers are presumed to be attributable to structural and stoichiometric variations within the films. Reported breakdown voltages for both the annealed and the as-deposited TiO2 oxide layers are in the range of 1 – 4 MV/cm, which is acceptable for current devices. The measured leakage currents of Ta2O5 films in the literature were around ;131027 A/cm2, with the exception of one datum of 1029 A/cm reported by Robert et al.38 Generally, Y2O3 layers show lower leakage current ~on the order of 10210 A/cm2! than other polycrystalline oxides. In any event, extensive work on both Ta2O5 and Y2O3 films after the initial publications did result in lower leakage currents. The leakage current of these biomimetic TiO2 films likewise needs to be lowered if device applications are to be realized. ACKNOWLEDGMENTS

The authors would like to thank Professor M. TabibAzar for his help with the C – V and I – V measurements and M. Stan for his help with the RBS measurements. Research funds from the Air Force office of Scientific Research are gratefully acknowledged. The manuscript was prepared while H.S. was an Alexander von Humboldt Research Fellow, and M.R.D. was a visiting scientist, at the Max-Planck Institute fu¨r Metallforschung in Stuttgart, Germany, which support is also gratefully acknowledged. J. Gosch, Electron. Design, Sept. ~1991!. N. Rausch and E. P. Burte, J. Electrochem. Soc. 140, 145 ~1993!. 3 P. A. Bertrand and P. D. Fleischauer, Thin Solid Films 103, 167 ~1983!. 4 S. B. Desu, Mater. Sci. Eng. B 13, 299 ~1992!. 5 A. E. Feuersanger, Proc. IEEE 52, 1463 ~1964!. 6 W. D. Brown and W. W. Grannemann, Solid-State Electron. 21, 838 ~1978!. 7 K. S. Yeung and Y. W. Lam, Thin Solid Films 109, 169 ~1983!. 8 T. Fuyuki and H. Matsunami, Jpn. J. Appl. Phys., Part 1 9, 1288 ~1986!. 9 K. Fukushima and I. Yamada, J. Appl. Phys. 65, 619 ~1988!. 10 N. Rausch and E. P. Burte, J. Electrochem. Soc. 140, 145 ~1993!. 11 M. Takeuchi, T. Itoh, and H. Nagasaka, Thin Solid Films 51, 83 ~1978!. 12 G. P. Burns, J. Appl. Phys. 65, 2095 ~1989!. 13 A. H. Heuer, D. J. Fink, V. J. Laraia, J. L. Arias, P. D. Calvert, K. Kendall, G. L. Messing, J. Blackwell, P. C. Rieke, D. H. Thompson, A. P. Wheeler, A. Veis, and A. I. Caplan, Science 225, 1098 ~1992!. 14 S. Mann, D. D. Archibald, J. M. Didymus, T. Douglas, B. R. Heywood, F. C. Meldrum, and N. J. Reeves, Science 261, 1286 ~1993!. 15 B. C. Bunker, P. C. Rieke, B. J. Tarasevich, A. A. Campbell, G. E. Fryx1 2

Shin, De Guire, and Heuer

3317

ell, G. L. Graff, L. Song, J. Liu, J. W. Virden, and G. L. McVay, Science 264, 48 ~1994!. 16 S. Baskaran, L. Song, J. Liu, Y. L. Chen, and G. L. Graff, J. Am. Ceram. Soc. ~submitted!. 17 ~a! A. Ulman, in An Introduction to Ultrathin Organic Films from Langmuir-Blodgett to Self-Assembly ~Academic, New York, 1991!; ~b! A. Ulman, Adv. Mater. 2, 573 ~1990!. 18 ~a! N. Balachander and C. N. Sukenik, Tetrahedron Lett. 29, 5593 ~1988!; ~b! Langmuir 6, 1621 ~1990!. 19 H. Shin, R. J. Collins, M. R. De Guire, A. H. Heuer, and C. N. Sukenik, J. Mater. Res. 10, 692 ~1995!. 20 M. R. De Guire, H. Shin, R. J. Collins, M. Agarwal, C. N. Sukenik, and A. H. Heuer, Proc. SPIE 2686, 88 ~1996!. 21 A. Rohatgi, S. R. Butler, F. J. Feigl, H. W. Kraner, and K. W. Jones, Appl. Phys. Lett. 30, 104 ~1977!. 22 B. J. Gordon, Solid State Technol. 57 ~1993!. 23 S. Wolf and R. N. Tauber, in Silicon Processing for the VLSI Era ~Lattice, Sunset Beach, CA, 1986!, Vol. 1, p. 199. 24 Z. H. Jin, D. V. Vezenov, Y. W. Lee, J. E. Zull, C. N. Sukenik, and R. F. Savinell, Langmuir 10, 2662 ~1994!. 25 H. Shin, Y. Wang, M. R. De Guire, and A. H. Heuer ~to be submitted!. 26 F. P. Fehlner, J. Electrochem. Soc. 119, 1723 ~1972!. 27 H. Shin, Ph.D. Dissertation, Case Western Reserve University, Cleveland, OH, 1996. 28 C. H. Ling, J. Bhaskaran, W. K. Choi, and L. K. Ah, J. Appl. Phys. 77, 6350 ~1995!. 29 L. M. Terman, Solid-State Electron. 5, 285 ~1962!. 30 E. H. Nicollian and J. R. Brews, MOS (Metal Oxide Semiconductor) Physics and Technology ~Wiley, New York, 1982!. 31 R. M. Anderson and D. R. Kerr, J. Appl. Phys. 48, 4834 ~1977!. 32 S.-I. Kimura, Y. Nishioka, A. Shinitani, and K. Mukai, J. Electrochem. Soc. 130, 2414 ~1983!. 33 A. C. Rastogi and R. N. Sharma, J. Appl. Phys. 71, 5041 ~1992!. 34 G. S. Oehrlein, J. Appl. Phys. 59, 1587 ~1986!. 35 M. Gurvitch, L. Manchanda, and J. M. Gibson, Appl. Phys. Lett. 51, 919 ~1987!. 36 L. Manchanda and M. Gurvitch, IEEE Electron Device Lett. 9, 180 ~1988!. 37 H. Shinriki and M. Nakata, IEEE Trans. Electron Devices 38, 455–62 ~1991!. 38 S. Roberts, J. Ryan, and L. Nesbit, J. Electrochem. Soc. 133, 1405 ~1986!. 39 A. Nagahori and R. Raj, J. Am. Ceram. Soc. 78, 1585 ~1995!. 40 I. Kim, J.-S. Kim, B.-W. Cho, S.-D. Ahn, J. S. Chun, and W.-J. Lee, J. Mater. Res. 10, 2864 ~1995!. 41 L. L. Hench and J. K. West, in Principles of Electronic Ceramics ~Wiley, New York, 1990!, p. 163. 42 E. Loh, J. Appl. Phys. 53, 6229 ~1982!. 43 J. M. Herbert, in Ceramic Dielectrics and Capacitors ~Gordon and Breach, New York, 1985!, p. 23. 44 N. Rausch and E. P. Burte, Microelectron. J. 25, 533 ~1994!. 45 R. Waser, T. Baiatu, and K.-H. Hardtl, J. Am. Ceram. Soc. 73, 1645 ~1990!. 46 R. Waser, J. Am. Ceram. Soc. 74, 1934 ~1991!. 47 Y. Nishioka, K. Kimura, H. Shinriki, and K. Mukai, J. Electrochem. Soc. 134, 410 ~1987!. 48 T. Oishi, in Chemical Processing of Ceramics, edited by B. I. Lee and E. J. A. Pope ~Dekker, New York, 1994!, pp. 465–479. 49 K.-I. Onisawa, M. Fuyama, K. Tamura, K. Taguchi, T. Nakayama, and Y. A. Ono, J. Appl. Phys. 68, 719 ~1990!. 50 W. M. Cranton, D. M. Spink, R. Stevens, and C. B. Thomas, Thin Solid Films 226, 156 ~1993!. 51 B. N. Sharma, S. T. Lakshmikumar, and A. C. Rastogi, Thin Solid Films 199, 1 ~1991!.