Tunneling-assisted Poole-Frenkel conduction mechanism in HfO2 thin films

Doo Seok Jeong and Cheol Seong Hwang Citation: J. Appl. Phys. 98, 113701 (2005); doi: 10.1063/1.2135895 View online: http://dx.doi.org/10.1063/1.2135895 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v98/i11 Published by the American Institute of Physics.

Additional information on J. Appl. Phys.

Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

JOURNAL OF APPLIED PHYSICS 98, 113701 2005

Tunneling-assisted Poole-Frenkel conduction mechanism in HfO2 thin lms

Doo Seok Jeong and Cheol Seong Hwanga
School of Materials Science and Engineering and Inter-university Semiconductor Research Center, Seoul National University, Seoul 151-742, Korea

Received 18 March 2005; accepted 14 October 2005; published online 2 December 2005 Tunneling-assisted Poole-Frenkel TAPF mechanism, which represents electron tunneling from a metal electrode to traps in a nearby insulator layer followed by detrapping of the electrons from the traps by virtue of a lowered potential well due to an applied electric eld, is suggested in this study to precisely describe the electrical conduction behavior of a Pt/ HfO2 / Si capacitor. The current density versus the applied electric-eld curves of the TAPF conduction show a similar electric-eld dependency to that of the Poole-Frenkel PF conduction. However, unlike the PF mechanism, the activation energy of the leakage current density corresponds to the value of the Schottky barrier height SBH of a metal/insulator junction minus a lowered potential-well height by the applied electric eld in the TAPF mechanism. In addition, the SBH of the Pt/ HfO2 junction is calculated considering a high space-charge density 1018 cm3 in the HfO2 layer. The measured activation energy for the electrical conduction from an atomic-layer-deposited HfO2 lm is compared to the value that is expected from this model. 2005 American Institute of Physics. DOI: 10.1063/1.2135895
I. INTRODUCTION

The Poole-Frenkel PF conduction mechanism is expected to be dominant in insulating thin lms when they have a large trap density and are thick enough to avoid quantum-mechanical tunneling.1 The high density of traps hinders the transport of electrons in the conduction band by drift and diffusion mechanisms and trapping and detrapping of electrons become the dominant processes that control the conduction-electron density in the insulating lms. In high-k dielectric constant thin-lm materials such as HfO2 and ZrO2, which bear a crucial importance as a gate dielectric layer for state-of-the-art semiconductor devices, the PF conduction behavior has been frequently observed due to the high density of traps.24 However, the PF conduction equation that has been used in many literatures is a somewhat simplied one which can give rise to inaccuracy. No consideration on the carrier injection at the electrode interface was included in the conventional PF equation the assumed carriers are electrons in this paper . In the PF mechanism, it is implicitly assumed that the carrier injection is much easier e.g., by accumulation contact formation so that the interface has very little inuence on the carrier concentration in the insulating layer. However, there are many thin-lm capacitor systems where the electrode/insulator interfaces comprise blocking contacts but still show the PF conduction mechanism when the lms are thick enough to suppress the direct tunneling and contain a rather high defect density. It can be easily imagined that the carrier concentration, and therefore the electrical conduction, can be signicantly inuenced by the contact properties carrier injection as well as the bulk trap properties in such cases. One obvious example that shows the inaccuracy of the
a

Electronic mail: cheolsh@plaza.snu.ac.kr

conventional PF equation for such a case is the unclear reason why the optical dielectric constant was obtained from the PF tting of the leakage-currentvoltage J-V data of the Pt/ HfO2 / Si capacitor when the electrons were injected from a Pt electrode.4 Since the Coulombic attraction between the positively charged trap center and conduction electron, that give rise to the PF effect, in the bulk of the lm is assumed to be a static phenomenon,5,6 a low-frequency k value was expected to be extracted from the PF tting. Recently, Jeong et al. gave a reasonable explanation for that by considering both the electron tunneling from a Pt electrode to the many traps in the vicinity of the interface of a metal/insulator junction and PF conduction.4 However, the classical PF equation itself cannot give an explicit description of the conduction behavior related to the detrapping of electrons assisted by the electric eld at the interface of the junction due to the error in the estimation of the conduction-electron density when the contacts are not so much injecting. Additionally, the number of electrons which can participate in the PF conduction is considered to be identical to that of the traps in the lm on the assumption that one trap donates one electron. And most of the traps are assumed to remain in electrically neutral states, i.e., the positively charged trap centers are always lled with electrons to maintain the neutral state. However, trapping and detrapping of electrons should be interpreted from the kinetic point of view because they take a certain specic time. Therefore, in this paper, the tunneling-assisted PooleFrenkel TAPF conduction mechanism was developed. The TAPF considers the effects on the carrier concentration of both the transition of electrons from an electrode to the bound states in the traps in the vicinity of the electrode by tunneling and detrapping of the electrons from the traps by virtue of thermal energy, and a lowered potential-well height by an externally applied electric eld. The remarkable difference of the TAPF conduction from the classical PF conduc 2005 American Institute of Physics

0021-8979/2005/98 11 /113701/6/$22.50

98, 113701-1

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

113701-2

D. S. Jeong and C. S. Hwang

J. Appl. Phys. 98, 113701 2005

tion is that the activation energy of the TAPF conduction must be the Schottky barrier height SBH of a metal/ insulator junction minus the potential-well height of the trap lowered by the electric eld. They show the same electriceld dependency. In addition, the inuence of the high trap density of the bulk insulator layer on the SBH of a metal/ insulator junction, that has been considered to be determined by work-function mismatch and Fermi-level pinning by interface states, was also calculated.
II. DERIVATION OF THE TUNNELING-ASSISTED POOLE-FRENKEL CONDUCTION EQUATION

tm in Eq. 1 can be obtained from the tunneling currentdensity j equation given by9 j=
m

E f m Et 1 f i Et f i Et 1 V 8 Et l
2

f m Et

dS jif

The electrons in the electrode are able to transit to the bound states originating from the traps in the insulator at the metal/insulator junction by quantum-mechanical tunneling. This has a high probability for the electrons of the electrode near the junction especially when the traps locate close to the interface. Therefore, the tunneling can serve as a signicant mechanism for supplying electrons to traps when the trap density is high, such as in the case of the Pt/ HfO2 interface. Successively, the trapped electrons can be thermally activated from the bound states of the traps to the conduction band of the insulator. This thermal activation becomes easier by virtue of the lowered potential-well height of the traps when an electric eld is applied PF effect . Hence, it is proper to regard the transition of the electrons from the electrode to the conduction band of the insulator via the traps, dened as the TAPF conduction, when an electric eld is applied to a system comprised of a blocking interface contact and an insulating layer with a high interfacial trap density. In order to derive the proper transport equation for the TAPF mechanism, the probabilities for the tunneling P1 to the bound state of the traps and the thermal activation P2 of the injected carriers should be calculated and multiplied. The tunneling probability of one electron p1 can be described using Fermis golden rule7 given by p1 = 2 tm
i f 2

where V, , l, f m Et , and f i Et are the volume of the electrode, the reduced Plancks constant, the thickness of the electrode, and the Fermi-Dirac distribution of electrons designating the occupancy of an energy state of Et in the electrode and insulator, respectively. m Et is the number of the allowed quantum states on the equal-energy surface in the reciprocal space of the electrode given by Eq. 11 of the present paper. From Eq. 4 the mean velocity of the tunneling electrons can be found considering the general ux equation multiplication of the density of particles and their mean velocity . Therefore, the transition time can be described as tm = Et 8 l
2

dS jif

It can be understood that the transition from the electrode to the traps requires more time as the physical distance between the two states and the difference between EF and bound-state energy Et increase. By entering Eq. 5 into Eq. 1 the absolute quantity of probability of the transition from the electrode to the traps can be obtained and described as a simple form of p1 = Et . 4l 6

where tm, , f , and i are the transition time, additional energy by perturbation, and the wave function of a nal bound state in the traps and an initial in the metal electrode states, respectively. For an easy calculation, the traps are assumed to be periodically distributed on a plane apart from the metal/insulator interface with a distance of and have an identical trap energy of Et, while the Fermi energy of the electrode EF is taken as reference. According to Bardeens theory, the matrix element i f can be substituted by the surface integrand of the probability currentdensity operator jif which is given by8
i f

It has to be noted that pl increases with increasing tm although the tunneling current density largely decreases with increasing tm. With increasing Et the decrease in the tunneling current is due to the large decrease in the number of tunneling electrons of the electrode, of which the energy must be the same as Et. As stated above, the traps are assumed to be periodically distributed on a plane distant from . the interface of the junction with the specic distance Therefore, to calculate the overall transition to many bound states the summation of the transition probabilities given by Eq. 6 should be carried out over all the bound states on the plane. Because it was assumed that all the bound states have an identical eigenvalue Et the summation can be simply described as P1 =
t

p1 =

N Et , 4l

=i

dS jif ,

where jif = Here,

* i

2im

* i

* i i.

is the conjugate of

where N is the number of the bound states on the plane. The probability of the transition of electrons from the bound states of the traps to the conduction band of the insulator can be described in terms of the lowered potential-well height by an electric eld and the thermal activation of electrons. This was already done in the classical PF conduction equation.1,5 Therefore, P2 can be given by

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

113701-3

D. S. Jeong and C. S. Hwang

J. Appl. Phys. 98, 113701 2005

P2 = exp

kT

where b, k, and T are the depth of the trap measured from the conduction-band edge of the insulator, the Boltzmann constant, and temperature, respectively. And b is given by1
b

q 3F s
r 0

1/2

where q, Fs, r, and 0 are the elementary charge, the electric eld at the interface of the metal/insulator junction, the optical dielectric constant of the dielectric, and the permittivity of vacuum, respectively. From the calculated P1 and P2 the probability of the transition of electrons from the electrode to the conduction band of the insulator via the trap bound states can be obtained by multiplying P1 by P2 since the transition can be regarded as a product of the two successive transitions. Furthermore, taking into account the number of allowed quantum states, which was quantied by Jeong et al.,9 the number of the electrons undergoing the transition is given by n= = where
m m

FIG. 1. Conduction-band prole of the Pt/ HfO2 / Si capacitor taking the interface layer between HfO2 and Si into consideration. The solid line is the prole drawn without considering ND and Qin/Si while the dashed line is the one considering ND and Qin/Si. The parameters for the conduction-band calculation are given in Table I.

Et f m Et P1 P2 Et Et + b N Et exp 4l kT
b

10

ent from the classical PF mechanism in terms of the barrier height although the barrier lowering is due to the Coulombic interaction between the positively charged trap center and conduction electron. More discussion about this point will be made in Sec. IV. The current density J of the TAPF conduction under a steady-state condition can be calculated using the general ux equation as J = qn
nF s ,

12

Et =

V
2

2meEt
2

3/2

2 2

1+

2meEtL2

1 ,

11

and f m Et is the Fermi-Dirac distribution function showing the occupancy of the energy state corresponding to Et in metal. me and L are the electron effective mass and the length of the longer edge of the electrode, respectively, that was assumed to be a thin slab-shaped parallelepiped as shown in Fig. 1 of Ref. 9. Electrons in the metal were assumed to be free electrons Eq. 11 . As a matter of fact, Et is a function of the electric eld in the insulator since the conduction-band prole of the insulator is changed by the electric eld. However, electrons can tunnel to the traps which locate very close to the interface so that only traps very close to the interface are concerned here. Therefore, the change in Et by an externally applied electric eld is negligible because the Fermi energy of the electrode is taken as a reference energy. From Eq. 10 , it can be noted that Et + b b corresponds to the activation energy of the TAPF mechanism. It should also be noted that Et + b corresponds to the SBH of the metal/insulator junction see Fig. 1 . In terms of the barrier height Et + b the TAPF mechanism appears to be the same as a thermionic emission Schottky emission mechanism where the carrier injection was mainly determined by the overcome of the barrier height at the interface with the help of the thermal energy. However, the TAPF mechanism is certainly different from the Schottky emission mechanism as b is given by Eq. 9 , i.e., the electric-eld dependency is different cf. in Schottky mecha3 1/2 due to the barrier lowering nism b = q Fs / 4 r 0 by image force effect . The TAPF mechanism is also differ-

where n is the electron mobility in the conduction band of the insulator. It has to be noted that Fs cannot be simply given as the VA / d, where VA and d are the applied voltage and insulating lm thickness, respectively, if the trap density is very high and the trap distribution throughout the thickness is not uniform.
III. CALCULATION OF THE SCHOTTKY BARRIER HEIGHT OF A METAL/INSULATOR JUNCTION IN A METAL/INSULATOR/SEMICONDUCTOR SYSTEM

The SBH of a metal-insulator MI contact with the presence of interface states was derived by Cowley and Sze and is given by10
b

=S

in

in

in

13

where m, in, and in are the work function of the metal, the charge neutral level of the insulator below the vacuum level, and the electron afnity of the insulator, respectively. S is the Schottky pinning parameter given by S= 1 1 + q D s/
2 id

14

where , id, and Ds are the thickness and permittivity of the interface dipole layer between the electrode and the insulator, and the density of the interface states per unit area and energy, respectively. Hobbs et al. reported Fermi-level pinning effect at the interface between polycrystalline silicon polySi and high-k gate oxide and the effects of the pinning on the atband voltage shifts. They proved that the atband voltage shift can mostly be attributed to the voltage drop by the interfacial dipole layer at the poly-Si gate and high-k gate

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

113701-4

D. S. Jeong and C. S. Hwang

J. Appl. Phys. 98, 113701 2005

TABLE I. Parameters for the conduction-band prole of the Pt/ HfO2 / Si capacitor shown in Fig. 1. HfO2 and IL are the permittivities of the HfO2 and the interface layer between HfO2 and Si, respectively. Parameter
m in in

Value 5.3 eV 4.8 eV 2.5 eV 0.53 4 18

Parameter
IL id

Value 8 1 1.2 1013 eV1 cm2 5 1018 cm3 2 1012 cm2

S
HfO2

Ds ND Qin/Si

oxide interface.11 Yeo et al. investigated the SBHs at the interfaces between various gate dielectrics and metal and poly-Si gates.12 Their experimental results showed a good agreement with the theoretical SBHs calculated by Eq. 13 . However, it should be noted that Eq. 13 was derived under the assumption that the bulk charge density of the insulating layer was negligible and there is no band bending in the insulator layer. The deposited high-k dielectric lms, either by chemicalvapor deposition CVD or atomic-layer deposition ALD usually have a rather high defect density and especially the region near the interface with the electrode is usually heavily damaged by the electrode fabrication process. This induces non-negligible space charges and band bending of the insulator layer, which should be considered for the calculation of SBH of the metal/insulator interface. As a matter of fact, a very high defect density 1018 cm3 in the deposited highk lms has been frequently reported.13,14 In addition, not only the metal/insulator interface and insulator bulk but also the insulator/Si interface has charges. Therefore, Eq. 13 should be modied to Eq. 15 given below by taking the insulator band bending and interface charges into consideration.

FIG. 2. Poole-Frenkel plot and the TAPF tting results for the Pt/ HfO2 / Si capacitor using the parameters given in Table I. The tting gives a barrier energy for electron conduction of 1.36 eV.

IV. EXPERIMENTAL RESULTS AND DISCUSSIONS

b=S

in

in

in

q2 NDtin + Qin/Si , 2 id + q Ds 15

where ND and tin are the positively charged defect density and thickness of the insulator, respectively. Qin/Si means charges per unit area at the interface between the insulator and Si. Here, positive interface charges are assumed, which has been suggested in various literatures where shifts of the atband voltages in the negative voltage direction are reported.15,16 For the derivation of Eq. 15 uniformly distributed space charges in the insulator are assumed. The detailed derivation procedure is given in the Appendix. It can be understood from Eq. 15 that SBH is also inuenced by ND and Qin/Si. The inuence of ND and Qin/Si on the SBH is shown in Fig. 1, where the calculated potential proles with and without ND and Qin/Si are shown with the other parameters for the calculation given in Table I for the Pt/ HfO2 / Si system. The SBH with and without the ND and Qin/Si is 2.32 and 2.57 eV, respectively.

For the experimental proving of the TAPF model for metal/high-k / Si systems, experimental J-Fs data from the authors previous reports for the Pt/ ALD-HfO2 16 nm thick /Si capacitors was taken.15 Details of the sample preparation and electrical measurement conditions were reported in Ref. 15. Figure 2 shows the experimental J-Fs data at measurement temperatures ranging from 30 to 135 C with the TAPF tting results lines . It can be noted that the TAPF mechanism can nicely t the data in the high eld region. The extracted barrier height was 1.36 eV. For the PF tting this value corresponds to the potential-well depth of the traps and for the TAPF this corresponds to the SBH of the Pt/ HfO2 interface. It seems that the TAPF does not hold its validity here considering the large difference between the calculated SBH 2.32 2.57 eV and extracted barrier height 1.36 eV . However, the simultaneously extracted optical dielectric constant4 suggests that the TAPF is a more appropriate conduction mechanism in this case. Although the authors ascribed 1.36 eV as the potential depth of traps in a HfO2 layer in Ref. 4 because the data were tted with the conventional PF equation, it is now believed that the TAPF is a more proper mechanism and the 1.36 eV should be interpreted as the SBH which is largely reduced by the factor described below. When the trap density reaches a very high value 1019 cm3 each trap begins to inuence the neighboring traps due to the very small distance between them. When the trap density is 1019 cm3 the average distance between the traps is only 4.6 nm. It is believed that the trap density near the interface is higher than that of the inner space of the insulator. The positively charged neighboring traps interfere with each other, and the potential well of the trap centers can be signicantly reduced under the applied bias. In an extreme case, the interfered Coulombic potential by the neighboring traps leads the elimination of the potential well to give rise to the transition from an electrical insulating state to a metallic state, which is the well-known Mott transition.17 Under such a circumstance, the potential-well depth of the traps b is largely reduced and the Et + b value extracted from the TAPF tting should be smaller than SBH, as seen in this case. The reason for obtaining the SBH as the barrier height of

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

113701-5

D. S. Jeong and C. S. Hwang

J. Appl. Phys. 98, 113701 2005

FIG. 3. Energy-band diagrams for the cases of a the bound state of Et located within the SBH and the high density of traps giving rise to the potential overlapping, b Et lower than EF and a high density of traps, and c Et lower than EF and a low density of traps which does not result in the potential overlapping. The schematic Fermi-Dirac distribution of electrons in the metal is also shown.

the TAPF mechanism when the reduced potential-well height is excluded can be understood from the following. Tunneling of electrons occurs only between energy states with the same energy. If the bound-state energy of the traps is higher than EF of metal, the electrons in the metal should be thermally activated to the trap energy level rst and then tunnel to the bound state thermionic eld emission . Then, the trapped electrons are thermally activated again from the bound-state energy to the conduction band of the insulator in order to conduct the electricity. With the help of the applied eld, this thermal activation process was largely enhanced by the PF effect and the large increase in conduction-electron density was obtained by increasing the applied voltage. Therefore, no matter whether the bound-state energy locates in the band gap, the barrier energy for the thermal activation should be Et + b as long as the trap energy location is higher than EF. Figure 3 a schematically shows this situation. The lowering of the trap potential-well depth by the overlapping of the modied potential between neighboring traps is also included. When the trap energy location is lower than EF and only one bound-state energy is allowed in the traps, which is usually the case due to the very small trap size,9 electrons in the metal of which the energy is lower than EF should tunnel to the bound states. Therefore, in this case, the SBH cannot be the barrier energy but the potential-well depth of the traps should be the barrier energy. This case actually corresponds to the conventional PF mechanism. Figure 3 b schematically shows this second situation which also shows the welldepth lowering by the potential overlapping. If the trap density is not as high so that the distance from the interface to the trap location is too large for tunneling, the electron injection to the traps may occur by thermionic emission. However, in this case, the activation energy for the thermionic emission SBH is far lower than the trap depth, so that the conduction behavior again follows the conventional PF mechanism. Figure 3 c schematically shows the third situation. In this case the well-depth lowering should not be considered.
V. CONCLUSION

interface. The rather limited injection of carriers through tunneling into the bound energy state of the traps makes a fundamentally different conduction mechanism from the conventional PF mechanism. Based on the fact that the conduction due to detrapping frequently takes place near the interface between the metal and insulator, the mathematical formula of the TAPF conduction mechanism was obtained by considering that the mechanism is a product event of two successive transition events: tunneling from the metal states to the bound states of the traps in the insulator and detrapping from the bound states to the conduction band of the insulator. Although the electric-eld dependence of the TAPF is the same as that of the conventional PF mechanism, the barrier energy obtained from the tting should correspond to the SBH whereas this corresponds to the potential-well depth in the conventional PF mechanism. If the potential-well depth is larger than the SBH, the conduction mechanism must follow the conventional PF mechanism. There was a large discrepancy between the experimentally extracted barrier height data 1.36 eV from the J-V behavior of the Pt/ HfO2 / Si system and the theoretically expected SBH 2.32 2.57 eV . This was considered to be due to the potential-well depth lowering by the overlapping of the potential proles between nearby traps. These traps converge under the applied bias voltage. It is believed that the TAPF mechanism better describe the electrical transport through an insulating lm where the contacts are of blocking nature and insulating lms contain a high trap density, such as noble-metal/high-k systems.

ACKNOWLEDGMENTS

The work was supported by the Korean Ministry of Science and Technology through the National Research Laboratories program, the National R&D Project for Nano Science and Technology M10214000097-02B1500-01500 , and the system IC 2010 program of the Korean government.

The TAPF conduction mechanism was developed in order to better understand the current conduction behavior of Pt/ HfO2 / Si capacitors. Since the interface potential barrier height was not negligible and the trap density was very high, the conventional PF mechanism cannot properly explain the current transport. It was incipiently assumed that the potential-well depth of the traps was lower than SBH of the

APPENDIX

The charge neutrality condition in a metal-insulatorsemiconductor capacitor is given by Eq. A1 considering the space charges in the insulator and the interface charges.

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

113701-6

D. S. Jeong and C. S. Hwang

J. Appl. Phys. 98, 113701 2005

Qm = NDtin + Qm/in + Qin/Si ,

A1

SBH equation given by Eq. 15 can be derived.

J. Frenkel, Phys. Rev. 54, 647 1938 . W. J. Zhu, T.-P. Ma, T. Tamagawa, J. Kim, and Y. Di, IEEE Electron Device Lett. 23, 97 2002 . 3 J. P. Chang and Y.-S. Lin, Appl. Phys. Lett. 79, 3666 2001 . 4 D. S. Jeong, H. B. Park, and C. S. Hwang, Appl. Phys. Lett. 86, 072903 2005 . 5 P. Hesto, in Instabilities in Silicon Devices, edited by G. Barbottin and A. Vapaille North-Holland, Amsterdam, The Netherlands, 1986 , p. 303. 6 N. F. Mott and R. W. Gurney, Electronic Processes in Ionic Crystals, 2nd ed. Clarendon, Oxford, 1940 , p. 85. 7 S. Gasiorowicz, Quantum Physics, 2nd ed. Wiley, New York, 1996 , p. 351. 8 J. Bardeen, Phys. Rev. Lett. 6, 57 1961 . 9 D. S. Jeong and C. S. Hwang, Phys. Rev. B 71, 016516 2005 . 10 A. M. Cowley and S. M. Sze, J. Appl. Phys. 36, 3212 1965 . 11 C. C. Hobbs et al., IEEE Trans. Electron Devices 51, 971 2004 ; 51, 978 2004 . 12 Y.-C. Yeo, P. Ranade, Q. Lu, R. Lin, T.-J. King, and C. Hu, Tech. Dig. VLSI Symp. 2001, 49. 13 D. S. Jeong, K. H. Ahn, W. Y. Park, and C. S. Hwang, Appl. Phys. Lett. 84, 94 2004 . 14 J. D. Baniecki, R. B. Laibowitz, T. M. Shaw, C. Parks, J. Lian, H. Xu, and Q. Y. Ma, J. Appl. Phys. 89, 2873 2001 . 15 H. B. Park, M. Cho, J. Park, C. S. Hwang, J.-C. Lee, and S.-J. Oh, J. Appl. Phys. 94, 1898 2003 . 16 H. B. Park, M. Cho, J. Park, S. W. Lee, T. J. Park, and C. S. Hwang, Electrochem. Solid-State Lett. 7, 252 2004 . 17 N. F. Mott, Proc. R. Soc. London, Ser. A 382, 1 1982 .
2 1

where Qm, ND, and tin denote the charges stored in the electrode, the positively charged defect density, and the thickness of the insulator, respectively. In Eq. A1 uniformly distributed space charges in the insulator are assumed. Qm/in and Qin/Si are charges per unit area at the interface between the metal and the insulator and between the insulator and Si, respectively. As mentioned in Ref. 10, the contribution of the space charges in Si to the charge neutrality is negligible. Then, the voltage drop Vid in the interface dipole layer can be calculated using Gauss law, which is given by Vid = where Qm/in = Ds
b

q
id

NDtin + Qm/in + Qin/Si ,

A2

in +

in

A3

Here, in and in are the charge neutral level of the insulator below the vacuum level and the electron afnity of the insulator, respectively. Equation A3 was derived by Cowley and Sze see Ref. 10 . Because Vid should be equal to Vid = m in + b q, Vid in Eq. A2 can be eliminated and by introducing the Schottky pinning parameter S the

Downloaded 13 Oct 2012 to 190.69.209.180. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions