G746
`
`Journal of The Electrochemical Society, 151 ~11! G746-G750 ~2004!
`0013-4651/2004/151~11!/G746/5/$7.00 © The Electrochemical Society, Inc.
`Failure Mechanism of Amorphous and Crystalline Ta-N Films
`in the Cu(cid:213)Ta-N(cid:213)Ta(cid:213)SiO2 Structure
`
`Ching-Chun Chang, J. S. Chen,a,*,z and Wu-Shiung Hsub
`aDepartment of Materials Science and Engineering, National Cheng Kung University, Tainan, Taiwan
`bNuclear Science Technology Development Center, National Tsing Hua University, Hsinchu, Taiwan
`
`The diffusion barrier properties of as-deposited amorphous TaNx (x ’ 0.5) and crystalline TaN between Cu and SiO2 have been
`investigated in Cu/Ta-N/Ta/SiO2 structures. The thermal reactions of Cu/TaNx /Ta/SiO2 and Cu/TaN/Ta/SiO2 after annealing in
`vacuum at 500 to 900°C were investigated by using sheet resistance measurements, glancing incident angle X-ray diffraction,
`scanning electron microscopy, energy-dispersive X-ray spectrometry, and Rutherford backscattering spectrometry. No significant
`reaction and change of sheet resistance were detected for both systems after annealing up to 800°C. As compared to TaN, TaNx
`exhibited better electrical properties and capability for preventing Cu diffusing through it. However, the sheet resistance of both
`systems increased abruptly after annealing at 900°C, especially the TaNx system. The severe increase in sheet resistance corre-
`sponds to the deterioration of Cu surfaces. Broken holes were seen in the TaNx layer, which were the initial sites for the structural
`failure. The cause of failure in Cu/Ta-N/Ta/SiO2 stacks is discussed on the basis of the characteristics of Ta-N films upon
`heat-treatment.
`© 2004 The Electrochemical Society.
`
`@DOI: 10.1149/1.1803836# All rights reserved.
`
`Manuscript received June 11, 2003. Available electronically October 7, 2004.
`
`With demands for an increase in the packing density and im-
`provement in device performance, the linewidths of integrated cir-
`cuits have reduced continuously to deep submicrometer dimensions.
`In ultralarge-scale integrated ~ULSI! circuits, resistance-capacitance
`~RC! time delay and electromigration become the important issues.
`Consequently, aluminum-based metallurgy is no larger adequate for
`deep submicrometer metallization. In order to solve these problems,
`copper has been adopted as the interconnection metal because of its
`lower resistivity ~1.67 mV cm! as compared with aluminum ~2.7 mV
`cm!. Meanwhile, the resistance to electromigration of copper is
`higher, also.1-3 However, copper diffuses easily into Si and SiO2 to
`form Cu-Si compounds at quite low temperatures.4,5 This causes
`device performance to degrade seriously. To avoid copper diffusion,
`a barrier between copper and its underlying dielectric layer is essen-
`tial. For copper metallization, diffusion barriers of refractory metals
`and their nitrides have been studied extensively owing to their su-
`perior thermal stability and high conductivity, including Ti-N,6,7
`Ta-N,8-12 and W-N.13-16 Among them, tantalum and its nitride draw
`lots of attention because they possess better thermal stability and
`chemical inertness than the other transition metal nitrides when
`coming into contact with copper.
`However, tantalum nitride may be in the form of TaN or Ta2N,
`and it can be crystalline or amorphous. Due to the different deposi-
`tion conditions, the properties of tantalum nitride barriers can vary
`widely.11,17-19 In the literature, most studies related with barrier per-
`formance concern only one tantalum nitride film of one specific
`composition and structure. In the present study, we deposited tanta-
`lum nitride film by reactive sputtering. Amorphous tantalum nitride
`films (TaNx , x ’ 0.5) and polycrystalline tantalum nitride films
`~TaN! were obtained by changing the sputtering ambient. In addi-
`tion, most of the literature reports concern the interactions of Cu
`films deposited on Ta-N/^Si& substrates. However, the interactions of
`Cu with the Ta-N/SiO2 /^Si& structure should be more relevant to the
`current Cu interconnect system. Therefore, we compared the reac-
`tions in the two types of Cu/Ta-N(TaNx or TaN!/Ta/SiO2/^Si& ~^Si&
`represents the single-crystal Si substrate! stacks after annealing at
`500 to 900°C.
`Sheet resistance, phases, elemental depth profiles, and surface
`morphology of the samples were examined. All these analyses could
`give us a guideline for the selection of the tantalum nitride as the
`diffusion barrier in the ULSI devices.
`
`* Electrochemical Society Active Member.
`z E-mail: jenschen@mail.ncku.edu.tw
`
`Experimental
`The substrates used in the present study were n-type ~100! Si
`wafers with resistivity of 1-10 V cm. The substrates were immersed
`in an organic bath and chemically etched with dilute HF solution
`(HF:H2O 5 1:10). Thermal SiO2 film, 280 nm in thickness, was
`grown by oxidizing Si wafers in dry oxygen at 1050°C. Ta-N films,
`50 nm in thickness, were deposited by radio frequency ~rf! sputter-
`ing from a Ta metal ~99.95% purity! target in different nitrogen-
`argon mixed ambients and applied with a negative substrate bias of
`2100 V. The films prepared with 1 and 5% of nitrogen flow ratio
`@N2 /(N2 1 Ar)# are amorphous TaNx (x ’ 0.5) and polycrystal-
`line TaN, respectively. More detailed information about the Ta-N
`films can be found in our previous work.20 Before Ta-N deposition,
`a 10 nm thick Ta layer was deposited to improve the adhesion be-
`tween Ta-N and SiO2 layer. Cu films ~180 nm! were then deposited
`on Ta-N films using dc sputtering with a Cu target ~99.99% purity!.
`The two groups of Cu/Ta-N/Ta/SiO2 /^Si& samples were then an-
`nealed side by side in vacuum (2.5 3 1025 Torr! at temperatures
`ranging from 500 to 900°C for 30 min to investigate thermal inter-
`actions. Sheet resistances of all samples, before and after annealing,
`were measured with a four-point probe. The crystalline structures of
`Cu/Ta-N/Ta/SiO2 /^Si& were characterized by using glancing inci-
`dent angle X-ray diffraction ~GIAXRD, Rigaku D/MAX2500! at an
`incident angle of 2° with Cu Ka radiation. Surface morphology of
`the films was examined by scanning electron microscopy ~SEM,
`Philips XL-40FEG!. The variations of surface compositions were
`estimated by energy dispersive X-ray spectrometry ~EDS, Philips
`EDAXDX-4!. Depth profile analysis was performed with Rutherford
`backscattering spectrometry ~RBS!. For RBS measurement,
`the
`4He1 ions were accelerated to 2 MeV and the backscattered ions
`were detected at a scattering angle of 160°.
`
`Results and Discussion
`The sheet resistances of all the samples, before and after anneal-
`ing, were characterized by a four-point probe and presented in Fig.
`1. The sheet resistance values of Cu/TaNx /Ta/SiO2 /^Si& samples
`were lower than those of Cu/TaN/Ta/SiO2 /^Si& samples upon an-
`nealing to 900°C, which might be due to the lower resistivity of
`TaNx ~;200 mV cm! than that of TaN ~;340 mV cm!, and fewer
`defects in the films before annealing at high temperature. In general,
`the measured sheet resistance was dominated by the copper film
`since the resistivity of copper is much lower than that of Ta-N film.
`Therefore, the variations of the measured sheet resistance may rep-
`resent the changes in the structure or composition of Cu film, or the
`intermixing degree of copper film with the underlayer.
`
`Downloaded 10 Sep 2008 to 140.116.208.41. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
`
`Page 1 of 5
`
`IP Bridge Exhibit 2001
`TSMC v. IP Bridge
`IPR2016-01249
`
`

`
`Journal of The Electrochemical Society, 151 ~11! G746-G750 ~2004!
`
`G747
`
`Figure 1. Dependence of sheet resistance of Cu/Ta-N/Ta/SiO2 /^Si& samples
`on the annealing temperatures.
`
`The sheet resistance values of both systems decreased with in-
`creasing annealing temperatures until 600°C. It is mainly attributed
`to copper grain growth and the self-healing effect of defects in the
`Cu film. After annealing at 700°C or above, however, the sheet
`resistance values rise slightly as the annealing temperature in-
`creases. After annealing at 900°C,
`the sheet resistance of the
`Cu/TaNx /Ta/SiO2 /^Si&
`sample increased dramatically to 1113
`mV/h and that of the Cu/TaN/Ta/SiO2 /^Si& sample increased to
`610 mV/h. To further understand the mechanisms that made the
`differences in these two systems, several material characterizations
`were carried out as follows.
`Figure 2 presents the GIAXRD patterns of Cu/TaNx /Ta/
`SiO2 /^Si& and Cu/TaN/Ta/SiO2 /^Si& samples before and after an-
`nealing at 500-900°C. Copper diffraction peaks and a broad peak at
`2u ’ 37° are seen in the as-deposited Cu/TaNx /Ta/SiO2 /^Si&
`sample ~Fig. 2a!, indicating that the TaNx layer is amorphous. The
`crystallization temperature of amorphous TaNx has been determined
`by annealing the film at temperatures from 300 to 900°C in 100°C
`intervals. This indicates that the TaNx layer began to crystallize into
`a Ta2N phase after annealing at 500°C. In Fig. 2b, only the diffrac-
`tion peaks of TaN and Cu were observed in the as-deposited and
`500°C-annealed samples. However, we can find the Ta2N(100) ~at
`2u ’ 34°) and Ta2N(101) ~at 2u ’ 39°) diffraction peaks in the
`pattern of the 600°C annealed sample. According to the composition
`obtained by RBS analysis, the TaN film on a graphite substrate
`showed the composition of Ta48N52 . Therefore, the existing Ta2N
`phase after annealing at 600°C is attributed to the reaction of the
`thin Ta underlayer with the excess nitrogen atoms in the TaN film.
`After annealing at 900°C, the intensities of Cu diffraction peaks
`decreased in both the Cu/Ta-N/Ta/SiO2 /^Si& systems, which might
`be attributed to the diffusion of copper into the underlayer and re-
`action with the silicon to form copper silicide.
`SEM micrographs on the surfaces of both Cu/TaNx /Ta/SiO2 /
`^Si& and Cu/TaN/Ta/SiO2 /^Si& samples after annealing at 700, 800,
`and 900°C are shown in Fig. 3 and 4. The surfaces of Cu/TaNx and
`Cu/TaN systems after annealing at 700°C simply show the mor-
`phology of Cu grains ~Fig. 3a and b!. After annealing at 800°C,
`broken holes are evident on the copper surfaces which appeared
`in the Cu/TaNx systems ~Fig. 3c!. However, there were only tiny
`voids observed in the Cu/TaN system ~Fig. 3d!. It is observed
`that Cu grains started to agglomerate after annealing at 700°C
`in this study. Moreover, holes and voids formed on the 800°C-
`annealed samples would increase the sheet
`resistance further.
`The rough surface morphology of Cu correlated well with the in-
`
`~a! Cu/TaNx /Ta/SiO2 /^Si&
`and ~b!
`Figure 2. GIAXRD patterns of
`Cu/TaN/Ta/SiO2 /^Si& samples as deposited and after annealing at 500, 600,
`700, 800, and 900°C.
`
`creases in the sheet resistance ~Fig. 1! in both Cu/TaNx and Cu/TaN
`systems.
`After annealing at 900°C, gray dots could be seen with the naked
`eye on the surface of the Cu/TaNx sample. The gray dots under SEM
`can be represented by the micrograph of Fig. 4a and the dark region
`at the center seems to be a broken area. On the other hand, the
`surface of the Cu/TaN system under SEM exhibited a circular region
`of different contrast but no broken area ~Fig. 4b!. Except for the
`gray spots,
`the surface morphology of the other areas on the
`Cu/TaNx system was basically similar to the Cu/TaN system. SEM
`analysis indicates that more severe local reaction occurred in the
`Cu/TaNx /Ta/SiO2 /^Si& stack than the Cu/TaN/Ta/SiO2 /^Si& stack
`after annealing at 900°C. Comparing the surface morphology of
`Cu/TaNx and Cu/TaN systems, we can conclude that there must be
`local defects originally existing in both TaNx and TaN films. Upon
`900°C annealing, the local defects in the TaNx film would change
`into apparent holes, but this phenomenon did not happen in the TaN
`film. The reasons for the diversity is discussed later.
`As for the circular spots on the surface of the 900°C-annealed
`Cu/TaNx system, they can be divided into three regions according to
`their distinct appearances, and labeled as region I, region II, and
`region III ~see Fig. 4a!. Figure 5 shows magnified micrographs of
`the 900°C-annealed Cu/TaNx /Ta/SiO2 /^Si& sample at region I, II,
`and III, respectively. EDS analysis was used to identify the chemical
`compositions of these regions. To reduce the inaccuracy of the com-
`positions influenced by the underlayer, the operating voltage of the
`electron beam was lowered to 10 kV so that the X-ray generation
`depth is less than 0.1 mm.
`
`Downloaded 10 Sep 2008 to 140.116.208.41. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
`
`Page 2 of 5
`
`

`
`G748
`
`Journal of The Electrochemical Society, 151 ~11! G746-G750 ~2004!
`
`Figure 4. SEM images of the surfaces of ~a! Cu/TaNx /Ta/SiO2 /^Si& and ~b!
`Cu/TaN/Ta/SiO2 /^Si& after annealing at 900°C.
`
`respectively. Therefore, the rough surface layer is Cu and the under-
`layer is TaNx . In region III, copper layer agglomerated but still
`remained continuous in surface morphology. EDS analysis was also
`used to detect the composition distributions on the Cu/TaN sample
`surface ~Fig. 4b! and the locations detected were labeled as a, b, and
`c in Fig. 6. Based on the EDS analysis, we have determined that the
`large grains and spherical clusters ~locations labeled as a and c,
`respectively! on the surface of the Cu/TaN samples after annealing
`at high temperatures consist of Cu and O. Meanwhile, the matrix
`revealed on the surfaces of Cu/TaN samples ~labeled as b! after
`annealing should be the TaN layer.
`As regards interfacial diffusion, compositional depth profiles of
`the as-deposited and 700°C-annealed samples were investigated by
`RBS and the spectra are shown in Fig. 7. The RBS spectra of
`samples annealing at higher annealing temperatures ~800 or 900°C!
`are not shown because the surface morphology of these samples is
`not uniform ~scattered with voids or spots! so that they are not
`appropriate for RBS analysis. Figure 7 shows that there is a small Ta
`signal present on the surface of the Cu/TaNx /Ta/SiO2 /^Si& sample
`after annealing at 700°C. The rest of the profile shifts slightly to the
`left due to the surface Ta. It had been reported that Ta has a very
`high affinity to oxygen and reacts with it to form Ta2O5 .21 Conse-
`quently, some Ta atoms may penetrate through the Cu layer to the
`surface to react with the residual oxygen in the annealing ambient.
`Except that, the Cu tail is a little slanting as compared to the as-
`deposited profile. It is said that as Ta diffused out to the surface,
`some vacancies would be left behind at the interface of Cu and
`TaNx . Therefore, Cu atoms could diffuse into the TaNx layer. Dif-
`fusion of Cu into the TaNx layer is only minuscule, as shown in Fig.
`7a. Furthermore, Ta diffusing through the extended defects, such as
`grain boundaries, would decorate and block the active paths for
`
`Figure 3. SEM images of the surfaces of ~a! Cu/TaNx /Ta/SiO2 /^Si&,
`and ~b! Cu/TaN/Ta/SiO2 /^Si&
`annealing at 700°C and ~c!
`after
`Cu/TaNx /Ta/SiO2 /^Si&, and ~d! Cu/TaN/Ta/SiO2 /^Si& after annealing
`at 800°C.
`
`The EDS spectrum shows that in region I ~Fig. 5a!, the large
`grains ~labeled a! mainly consist of Cu. The major elements in the
`region labeled b are copper, silicon, and oxygen, indicating that the
`matrix in region I is the SiO2 layer. This means that the top Cu layer
`and the underlaying TaNx film had seriously deteriorated, therefore
`the SiO2 layer was revealed in region I. The EDS spectrum in region
`II ~area labeled c! consists of tantalum and oxygen ~Fig. 5b!. The
`low content of copper suggests that the grainy matrix observed in
`region II is the TaNx film, and it may be oxidized. The TaNx under-
`layer was revealed because the Cu top layer delaminated seriously
`after annealing at high temperature. At region III ~Fig. 5c!, the de-
`tected elements are Cu and Ta, for the area labeled as d and e,
`
`Downloaded 10 Sep 2008 to 140.116.208.41. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
`
`Page 3 of 5
`
`

`
`Journal of The Electrochemical Society, 151 ~11! G746-G750 ~2004!
`
`G749
`
`Figure 6. Magnified SEM images of the surface of Cu/TaN/Ta/SiO2 /^Si&
`after annealing at 900°C.
`
`20.5 nm after annealing at 900°C. However, the crystallite size of
`the 900°C-annealed polycrystalline TaN film was 14.2 nm, which
`was almost the same as that of the as-deposited TaN film ~12.4 nm!.
`In addition, Ta2N has a melting point of 2050°C, compared with that
`of 3087°C for TaN. According to the empirical relationship, the
`activation energy of grain growth of Ta2N is projected to be signifi-
`cantly lower than its TaN counterpart. Therefore, grain growth of
`Ta2N is expected to be in evidence. According to Chaudhari’s report,
`
`Figure 5. Magnified SEM images of the surfaces of ~a! region I, ~b! region
`II, and ~c! region III in Cu/TaNx /Ta/SiO2 /^Si& after annealing at 900°C.
`
`grain boundary diffusion. It will improve the capability to inhibit Cu
`from diffusing. On the other hand, no surface Ta signal is found in
`the RBS spectrum of the 700°C-annealed Cu/TaN/Ta/SiO2 /^Si&
`sample, while the Ta profile becomes wide-spreading ~Fig. 7b!.
`Also, the copper tail becomes slanting, in a slightly higher degree
`than the previous system. The RBS profile indicates that interdiffu-
`sion occurred at the Cu/TaN and TaN/Ta/SiO2 interfaces for the
`700°C annealed Cu/TaN/Ta/SiO2 /^Si& sample.
`Crystallite sizes of TaNx and TaN films were estimated by using
`the Scherrer equation22 and presented in Table I. The crystallite sizes
`were calculated from the full-width at half maximum ~fwhm! of the
`Ta2N ~101! peak and TaN ~111! peak, respectively. The table indi-
`cates that the crystallite size of the TaNx , which underwent the
`amorphous-to-crystalline transformation,
`increased drastically to
`
`the ~a! Cu/TaNx /Ta/SiO2 /^Si&
`and ~b!
`Figure 7. RBS spectra of
`Cu/TaN/Ta/SiO2 /^Si& samples, as deposited and after annealing at 700°C.
`The arrows indicate the backscattered energies of Cu and Ta on sample
`surface.
`
`Downloaded 10 Sep 2008 to 140.116.208.41. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
`
`Page 4 of 5
`
`

`
`G750
`
`Journal of The Electrochemical Society, 151 ~11! G746-G750 ~2004!
`
`Table I. Estimated crystallite sizes of Ta-N films.
`
`As-deposited
`
`After annealing at 900°C
`
`Ta2N
`~nm!
`fl
`20.5
`
`TaN
`~nm!
`
`12.4
`
`14.2
`
`tensile stress will be produced in film as grains grow.23 Furthermore,
`Chuang et al.24 had reported that voids would be formed on the
`surfaces of Ta or Ta-N film to relieve the surface tension. In our
`study, the crystallite size of TaNx apparently increased after anneal-
`ing. Consequently, the as-deposited amorphous TaNx film, which
`crystallized into a Ta2N phase and underwent huge grain growth,
`must be stressed severely. To relieve the additional tensile stress,
`cracks or voids were formed. On the other hand, TaN film main-
`tained similar crystallite size to the as-deposited film even after an-
`nealing at high temperature. This indicates that the TaN film sus-
`tained less tensile stress than the Ta2N film did after annealing.
`Therefore, there were no apparent cracks and voids observed on the
`TaN film.
`Now, we can infer a mechanism for the broken holes formed in
`the Cu/TaNx system. At first, there may be some local defects exist-
`ing in the thermally grown SiO2 film, which could act as fast diffu-
`sion paths for copper to penetrate and then react with the Si sub-
`strate. However,
`the amorphous TaNx diffusion barrier between
`copper and SiO2 could prevent Cu from diffusing through it effi-
`ciently. After annealing at high temperatures, amorphous TaNx film
`tended to crystallize into Ta2N phases and was stressed severely so
`that lots of grain boundaries and voids formed. Consequently, the
`copper upper layer penetrated through the grain boundaries and
`voids in the TaNx , and the defects in the SiO2 film to the Si sub-
`strate and reacted with it. The broken holes were also observed by
`Holloway et al.9 using Ta and Tsai et al.17 using TaN. The differ-
`ence between our experiments and their studies is that there is a
`SiO2 layer between the diffusion barrier and the silicon substrate
`in this study. According to Tsai et al., the gray dots might be the
`initial sites on the TaN film for copper to penetrate. In the present
`study, it indicated that the conditions for copper diffusing were not
`only the grain boundaries or defects in the barrier but also the SiO2
`underlayer.
`The morphology of Cu in region II ~Fig. 5b! was circular dots
`scattered on the TaNx matrix. According to Miller et al.,25 one of the
`important factors in determining the possibility of agglomeration is
`the ratio of the film thickness to the grain size. When the grain-size-
`to-film-thickness ratio exceeds a critical value, the breakup will
`lower the free energy of the system. Hence, copper tended to ag-
`glomerate seriously around the reaction spots because the process of
`copper penetrating through the defects in the center of the reaction
`spots leads to the copper layer becoming thinner and thinner. There-
`fore, the grain-size-to-film-thickness ratio in region II should be
`larger than it was in region III. This means that the copper layer in
`region II underwent a more serious agglomeration than region III
`and became round-shaped clusters as shown in Fig. 5b.
`On the contrary, the Cu/TaN system still keeps the microstructure
`intact owing to minor grain growth even after annealing at 900°C.
`Therefore, no broken holes were observed on the surface of
`Cu/TaN/Ta/SiO2 /^Si& samples.
`Conclusion
`The criteria to choose an appropriate diffusion barrier include not
`only low resistivity but also excellent integrity and capability for
`
`preventing copper from penetrating through it. Cu/TaNx ~amor-
`phous, x ’ 0.5)/Ta/SiO2 /^Si& samples possessed lower sheet resis-
`tances than Cu/TaN~crystalline!/Ta/SiO2 /^Si& samples until anneal-
`ing at 800°C. After annealing at 900°C, the sheet resistance value of
`Cu/TaNx /Ta/SiO2 /^Si& sample increased drastically, which was
`about twice as large as that of Cu/TaN/Ta/SiO2 /^Si&. By using SEM
`and EDS analyses, we found that the TaNx film between Cu and
`SiO2 bears an additional tensile stress because of its substantial
`grain growth, which makes voids produced in the TaNx film after
`annealing at 900°C. These defects existing in the TaNx film then
`resulted in apparent reaction spots on the surface of Cu. On the other
`hand, TaN grains did not grow apparently so that the TaN barrier can
`keep its integrity even after annealing at 900°C. However, at lower
`annealing temperatures, RBS spectra indicate that the TaNx film has
`a better ability to prevent the copper diffusion due to the amorphous
`character of TaNx . Therefore, we may conclude that the TaNx film is
`more appropriate than the TaN film for applications in integrated
`circuits. Nevertheless, the TaNx film degrades seriously after anneal-
`ing at 900°C, indicating that TaNx may be less sustainable than TaN
`when encountering a severe upsurge in temperature.
`
`Acknowledgments
`The authors gratefully acknowledge the financial support from
`the National Science Council of Taiwan ~grant no. NSC-91-2216-E-
`006-059!.
`
`National Cheng Kung University assisted in meeting the publication
`costs of this article.
`
`References
`1. C. W. Park and R. W. Vook, Appl. Phys. Lett., 59, 175 ~1991!.
`2. P. J. Pan and C. H. Ting, IEEE Trans. Ind. Electron., IE-29, 154 ~1982!.
`3. S. P. Murarka, Mater. Sci. Eng., R., 19, 87 ~1997!.
`4. C. S. Liu and L. J. Chen, J. Appl. Phys., 74, 5501 ~1993!.
`5. A. Cros, M. O. Aboelfotoh, and K. N. Tu, J. Appl. Phys., 67, 3328 ~1990!.
`6. S. Q. Wang, I. J. M. M. Raaijmakers, B. J. Burrow, S. Suthar, S. Redkar, and K. B.
`Kim, J. Appl. Phys., 68, 5176 ~1990!.
`7. K. C. Park, S. H. Kim, and K. B. Kim, J. Electrochem. Soc., 147, 2711 ~2000!.
`8. E. Kolawa, J. S. Chen, J. S. Reid, P. J. Pokela, and M. A. Nicolet, J. Appl. Phys.,
`70, 1369 ~1991!.
`9. K. Holloway, P. M. Fryer, C. Cabral, Jr., J. M. E. Harper, P. J. Bailey, and K. H.
`Kelleher, J. Appl. Phys., 71, 5433 ~1992!.
`10. K. H. Min, K. C. Chun, and K. B. Kim, J. Vac. Sci. Technol. B, 14, 3263 ~1996!.
`11. G. S. Chen and S. T. Chen, J. Appl. Phys., 87, 8473 ~2000!.
`12. K. M. Latt, Y. K. Lee, S. Li, T. Osopowicz, and H. L. Seng, Mater. Sci. Eng., B, 84,
`217 ~2001!.
`13. K. M. Chang, T. H. Yeh, I. C. Deng, and C. W. Shin, J. Appl. Phys., 82, 1469
`~1997!.
`14. J. E. Kelsey, C. Goldberg, G. Nuesca, G. Peterson, A. E. Kaloyeros, and B. Arkles,
`J. Vac. Sci. Technol. B, 17, 1101 ~1999!.
`15. A. E. Kaloyeros and E. Eisenbraun, Annu. Rev. Mater. Sci., 30, 363 ~2000!.
`16. B. M. Ekstrom, S. Lee, N. Magtoto, and J. A. Kelber, Appl. Surf. Sci., 171, 275
`~2001!.
`17. M. H. Tsai, S. C. Sun, C. E. Tsai, S. H. Chuang, and H. T. Chiu, J. Appl. Phys., 79,
`6932 ~1996!.
`18. Y. K. Lee, K. M. Latt, J. H. Kim, T. Osipowicz, S. Y. Chaim, and K. S. Lee, Mater.
`Sci. Eng., B, 77, 282 ~2000!.
`19. H. Wang, A. Tiwari, X. Zhang, A. Kvit, and J. Narayan, Appl. Phys. Lett., 81, 1453
`~2002!.
`20. C. C. Chang, J. S. Jeng, and J. S. Chen, Thin Solid Films, 413, 46 ~2002!.
`21. K. Holloway and P. M. Fryer, Appl. Phys. Lett., 57, 1736 ~1990!.
`22. B. D. Cullity and S. R. Stock, Elements of X-Ray Diffraction, 3rd ed., p. 170,
`Prentice Hall, New Jersey ~2001!.
`23. P. Chaudhari, J. Vac. Sci. Technol., 9, 520 ~1971!.
`24. J. C. Chuang and M. C. Chen, Thin Solid Films, 322, 213 ~1998!.
`25. K. T. Miller, F. F. Lange, and D. B. Marshall, J. Mater. Res., 5, 151 ~1990!.
`
`Downloaded 10 Sep 2008 to 140.116.208.41. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp
`
`Page 5 of 5