Comparative study of tantalum and tantalum nitrides (Ta2N and TaN) as a diffusion
`barrier for Cu metallization
`Kyung‐Hoon Min, Kyu‐Chang Chun, and Ki‐Bum Kim
`
`Citation: 14, 3263 (1996); doi: 10.1116/1.588818
`View online: http://dx.doi.org/10.1116/1.588818
`View Table of Contents: http://avs.scitation.org/toc/jvn/14/5
`Published by the American Institute of Physics
`
`Page 1 of 8
`
`IP Bridge Exhibit 2042
`TSMC v. IP Bridge
`IPR2016-01264
`
`

`

`Comparative study of tantalum and tantalum nitrides (Ta2N and TaN)
`as a diffusion barrier for Cu metallization*
`Kyung-Hoon Min, Kyu-Chang Chun, and Ki-Bum Kim
`Department of Metallurgical Engineering, Inter-university Semiconductor Research Center, Seoul National
`University, Seoul, Korea
`共Received 7 November 1995; accepted 13 July 1996兲
`Tantalum 共Ta兲 and tantalum nitride films 共Ta2N and TaN兲 of about 50 nm thickness were reactively
`sputter deposited onto 共100兲 Si substrate by using dc magnetron sputtering and their diffusion barrier
`properties in between Cu and Si were investigated by using sheet resistance measurement, x-ray
`diffraction, Auger electron spectroscopy, and Secco etching. With increasing amounts of nitrogen in
`the sputtering gas, the phases in the as-deposited film have been identified as a mixture of ␤-Ta and
`bcc-Ta, bcc-Ta, amorphous Ta2N, and crystalline fcc-TaN. Diffusion barrier tests indicate that there
`are two competing mechanisms for the barrier failure; one is the migration of Cu into the Si
`substrate and another is the interfacial reaction between the barrier layer and the Si substrate. For
`instance, we identified that elemental Ta barrier failure occurs initially by the diffusion of Cu into
`the Si substrate through the barrier layer at 500 °C. On the other hand, the Ta2N barrier fails at
`700 °C by the interfacial reaction between Ta2N and Si substrate instead of the migration of Cu into
`the Si substrate. For the case of TaN, the barrier failure occurs by the migration of Cu into the Si
`substrate at 750 °C. It is also demonstrated that the diffusion barrier property is enhanced as the
`nitrogen concentration in the film is increased. © 1996 American Vacuum Society.
`
`I. INTRODUCTION
`
`Copper has drawn much attention as a new interconnect
`material for deep submicron integrated circuits 共ICs兲 as a
`replacement for Al and its alloys. The major motivation for
`this replacement is due to the lower resistivity and superior
`electromigration and stress migration resistance of Cu as
`compared to Al and its alloys.1,2 However, in order to suc-
`cessfully integrate Cu metallization into ICs, some problems
`associated with the transition to Cu such as lack of an aniso-
`tropic etching, oxidation, corrosion, and poor adhesion to
`most of the dielectric layers should be resolved. In particular,
`the diffusion of Cu into either Si and SiO2 layers should be
`retarded by employing a suitable diffusion barrier layer.3,4
`Indeed, there has been a considerable effort to identify a
`suitable diffusion barrier layer for Cu metallization. Materi-
`als investigated include Ta,5–8 W,9 TiW,10 TiSi2,11 TiN,12,13
`Ta2N,5 W2N,14 Ni0.6Nb0.4,15 and amorphous Ta–Si–N.8 The
`results of these efforts has been well summarized recently by
`Wang.16 Among these materials, tantalum has been exten-
`sively investigated as a diffusion barrier for Cu since it not
`only shows relatively high melting temperature but is also
`known to be thermodynamically stable with respect to Cu.17
`For instance, Holloway et al.5 and Catania et al.6 investi-
`gated the barrier properties of sputter deposited 50-nm-thick
`Ta layer and identified that the layer was stable up to 550
`and 650 °C, respectively. In contrast, Chang7 observed inter-
`mixing of Cu and Si through the Ta barrier layer even at
`300 °C by Rutherford backscattering spectroscopy 共RBS兲. In
`addition, Kolawa et al.8 identified that the junctions covered
`with a 180-nm-thick Ta barrier layer and Cu failed after an-
`nealing at 500 °C for 30 min. Thus, it appears that there still
`is a controversy about the barrier failure temperature of Ta.
`
`*Published without author corrections
`
`In particular, we note that the barrier failure temperature is
`quite different depending on the method to identify the bar-
`rier failure. In addition, we note that there are only few re-
`ports describing the barrier properties of tantalum nitrides
`such as Ta2N and TaN. Holloway et al.5 reported that 50-
`nm-thick Ta2N was stable up to 650 °C, thus indicating that
`the barrier property was improved by about 100 °C com-
`pared to that of pure Ta film. However, as far as we are
`aware of, the effectiveness of TaN as a diffusion barrier be-
`tween Cu and Si has not been reported so far.
`In this article, we would like to systematically investigate
`the diffusion barrier properties of Ta and its nitrides, both
`Ta2N and TaN, for Cu metallization and identify the barrier
`failure mechanism in each cases. In order to do this, we first
`reactively sputter deposited the films at various N2/Ar ratios
`and identified the phases and microstructure of the as-
`deposited film. Then, the barrier properties of the Ta, Ta2N,
`and TaN films were tested.
`
`II. EXPERIMENTS
`Tantalum and tantalum nitride films of about 50 nm thick-
`ness were deposited onto 共100兲 Si substrates by using dc
`magnetron sputtering at various N2/Ar gas ratios. Si wafers
`were cleaned in 10:1 diluted HF solution and rinsed in deion-
`ized water before loading into the chamber. During deposi-
`tion, the operating pressure was maintained at 10 mTorr and
`the substrates were water cooled. The phases and microstruc-
`tures of the as-deposited films were investigated by x-ray
`diffractometry 共XRD兲 and plan-view transmission electron
`microscopy 共plan-view TEM兲 operated at 200 kV and the
`nitrogen content of the as-deposited films was obtained by
`using RBS and Auger electron spectroscopy 共AES兲.
`In order to identify the barrier properties, 300-nm-thick
`Cu layer was deposited on top of the barrier layer without
`
`3263
`
`J. Vac. Sci. Technol. B 14(5), Sep/Oct 1996
`
`0734-211X/96/14(5)/3263/7/$10.00
`
`©1996 American Vacuum Society
`
`3263
`
`Page 2 of 8
`
`

`

`3264
`
`Min, Chun, and Kim: Tantalum and tantalum nitrides
`
`3264
`
`FIG. 1. The electrical resistivity and the nitrogen content of the films depos-
`ited at various N2/Ar gas ratios.
`
`FIG. 2. XRD patterns of the as-deposited Ta–N films deposited at various
`N2/Ar flow ratios. The numbers inside of the parentheses indicated the ni-
`trogen content in the film measured by RBS.
`
`breaking the vacuum and the samples were annealed for one
`hour at the temperatures ranging from 400 to 750 °C in hy-
`drogen ambient. Sheet resistance of the samples were mea-
`sured both before and after annealing by four-point probe to
`survey the overall reaction involving Cu. XRD and AES
`were used for the analysis of reaction product phases and the
`interdiffusion of the elements across the interface, respec-
`tively. Finally, Secco etching of the substrates was per-
`formed to identify the Cu penetration into the Si substrate
`after removing both Cu and barrier layers by wet-chemical
`solution.18,19 The wet-chemicals used in this experiment are
`HNO3:H2O⫽1:20 for Cu, HF:H2O⫽1:10 for Ta, and
`H2SO4:HF⫽9:1 for both Ta2N and TaN. The Si substrates
`were then Secco etched and examined by using optical mi-
`croscopy.
`
`III. EXPERIMENTAL RESULTS AND DISCUSSION
`
`A. The microstructure and phases of the as-
`deposited films
`
`Both electrical resistivity and nitrogen content of the films
`deposited at various N2/Ar gas ratios are shown in Fig. 1. It
`shows that the nitrogen content in the film is gradually in-
`creased as the partial flow of nitrogen in the sputtering gas is
`increased. The resistivity of the as-deposited film, however,
`shows several interesting features. It first shows that the elec-
`trical resistivity of the pure Ta film is about 150 ␮⍀ cm and
`is initially decreased to about 80 ␮⍀ cm as small amount of
`nitrogen is added to the sputtering gas. Then, the value of the
`resistivity gradually increased up to about 220 ␮⍀ cm as the
`nitrogen content in the film is increased to about 24 at. %. In
`between the nitrogen content of about 24 to 48 at. %, the
`resistivity of the film is only slightly increased from about
`220 to about 260 ␮⍀ cm. Finally, when further nitrogen is
`incorporated, the resistivity of the film drastically increased.
`The oxygen concentrations of pure Ta film, 24 at. % N con-
`tained film, and 48 at. % N contained film are almost the
`same. All the films contain about 1–2 at. % oxygen.
`In order to clearly understand the relationship between the
`resistivity and the nitrogen content in the film, both the
`
`J. Vac. Sci. Technol. B, Vol. 14, No. 5, Sep/Oct 1996
`
`phase共s兲 and the microstructure of the films have been inves-
`tigated by using XRD and plan-view TEM. First, Fig. 2
`shows the XRD patterns of Ta–N films containing various
`amount of nitrogen. The XRD peaks of the pure Ta film can
`be indexed as 共002兲 and 共004兲 of ␤-Ta except the one ap-
`pearing at 38°, which can also be indexed as 共202兲 of ␤-Ta or
`共110兲 of bcc-Ta. Thus, it is not clear from XRD whether the
`film only contains ␤-Ta or is a mixture of bcc-Ta and ␤-Ta.
`The XRD pattern of Ta共N兲 film deposited with 3% partial N2
`shows only one peak at 38°, which can be ascribed to either
`共110兲 of bcc-Ta or 共202兲 of ␤-Ta. These results indicate that
`the addition of a small amount of N2 in the sputtering gas
`either induce a phase transformation from ␤-Ta to bcc-Ta or
`induce the texture of the ␤-Ta film to change. For the films
`with higher nitrogen content, it is difficult to identify the
`phase共s兲 of the film by using XRD since only one or two
`weak and broad peaks appear.
`To further determine the phase共s兲 in the as-deposited
`films, the films were analyzed using plan-view TEM. Figure
`3 shows a series of bright field images and selected area
`diffraction 共SAD兲 patterns of the as-deposited films with dif-
`ferent amounts of nitrogen contents in the film. First, the
`bright field image and SAD of the pure Ta film clearly show
`that the film is composed of a mixture of ␤-Ta and bcc-Ta,
`with a grain size of about 20 to 30 nm 关Fig. 3共a兲兴. Thus, from
`the results of XRD and TEM, we can conclude that the pure
`Ta film consists of a mixture of ␤-Ta and bcc-Ta. Moreover,
`the ␤-Ta phase in the film is observed to form a strong 共002兲
`texture, while the bcc-Ta phase in the film forms a 共110兲
`texture. Figure 3共b兲 shows the bright field image and SAD
`pattern of the films deposited with 3% of N2 in the sputtering
`gas. The SAD pattern of this sample clearly shows that the
`film only contains bcc-Ta and the bright field image shows
`that the grain size of this film is similar to that of pure Ta
`film. Figure 3共c兲 shows both the TEM image and SAD pat-
`tern of the Ta–N film with 33 at. % of nitrogen in the film
`which shows that the film forms an amorphous phase. Fi-
`nally, Fig. 3共d兲 is the bright field image and SAD of the TaN
`film showing the formation of a crystalline fcc phase with
`
`Page 3 of 8
`
`

`

`3265
`
`Min, Chun, and Kim: Tantalum and tantalum nitrides
`
`3265
`
`FIG. 3. Plan-view TEM micrographs and selected area diffraction patterns of the as-deposited films: 共a兲 pure-Ta; 共b兲 Ta共N兲; 共c兲 Ta2N; and 共d兲 TaN.
`
`turn our attention to the diffusion barrier properties of each
`of these films. Figure 4 shows the variation of the sheet
`resistance of the samples upon annealing. The data mainly
`show changes in the thickness or resistivity of the unreacted
`Cu layer, since the sheet resistance of the barrier layer and
`reaction products are expected to be much larger than that of
`Cu. We first note that the sheet resistance of the film stack
`initially drops by annealing which apparently is caused by a
`decrease in defect density and grain growth in the Cu film.
`The sheet resistance of the Cu/Ta/Si film increased slightly
`
`grains having sizes of a few nm. Interplanar spacings derived
`from the SAD pattern agree with those of TaN.
`If one reviews the results of resistivity and phase identi-
`fication, the initial decrease of the resistivity from about 150
`␮⍀ cm to about 80 ␮⍀ cm with small N2 addition can be
`ascribed to the phase transformation from ␤-Ta to bcc-Ta.
`Indeed, it has been well known the typical reported resistiv-
`ities of ␤-Ta are about 180 ␮⍀ cm, and for bcc-Ta about 40
`␮⍀ cm.20,21 Although it is not clear yet why this phase trans-
`formation occurs by the addition of small amounts of nitro-
`gen in the sputtering gas, similar behavior has been reported
`by others.18,20–25 It is also interesting to note that an amor-
`phous phase is formed at about 33 at. % nitrogen content
`关Fig. 3共c兲兴. Reid et al.26 reported the formation of a mixture
`of amorphous and crystalline Ta2N phase close to this com-
`position. In contrast, Holloway et al.5 reported the formation
`of a crystalline Ta2N phase. The formation of a mixture of
`amorphous and crystalline Ta2N phase is also observed in
`our case. Therefore, at somewhat smaller concentration of
`nitrogen, it appears that the formation of amorphous, crystal-
`line, or mixtures of amorphous and crystalline Ta2N phase is
`all possible. It is believed that small variations of N content
`or the sputtering parameters can explain the different obser-
`vations.
`
`B. Barrier properties
`
`Having discussed the evolution of microstructure and
`phase共s兲 of the film by varying the nitrogen content, we now
`
`FIG. 4. Sheet resistance variation of the Cu/Ta/Si, Cu/amorphous Ta2N/Si,
`and Cu/TaN/Si samples as a function of annealing temperature.
`
`JVST B - Microelectronics and Nanometer Structures
`
`Page 4 of 8
`
`

`

`3266
`
`Min, Chun, and Kim: Tantalum and tantalum nitrides
`
`3266
`
`FIG. 5. XRD patterns of the Cu/Ta/Si sample after annealing at various
`temperatures: 共a兲 as-deposited; 共b兲 600 °C; and 共c兲 650 °C.
`
`FIG. 7. XRD patterns of the Cu/crystalline fcc-TaN/Si sample as a function
`of annealing temperature: 共a兲 as-deposited; 共b兲 650 °C; and 共c兲 750 °C.
`
`controversy on the role of Cu in crystallization process of
`some amorphous films. For instance, Reid et al.26 suggested
`that Cu reduces the crystallization temperature of some
`amorphous films. However, Thomas et al.28 reported that a
`
`upon annealing at 600 °C. However, after annealing at
`650 °C, the color of the sample is observed to change from
`Cu color to gray, and the sheet resistance of the sample can
`be observed to drastically increase, which indicates that a
`significant reaction has now occurred in between the layers.
`A similar behavior occurs at 700 °C for Cu/Ta2N/Si samples
`while no such behavior occurs for the Cu/TaN/Si samples
`even after annealing at 750 °C.
`Figure 5 shows the XRD results of the Cu/Ta/Si sample
`after annealing. It clearly shows the formation of Cu3Si at
`600 °C and the formation of Cu3Si and TaSi2 at 650 °C.
`These results are similar to those of Holloway et al.5 who
`also identified the abrupt increase of the sheet resistance with
`the formation of Cu3Si and TaSi2. XRD results of
`Cu/Ta2N/Si samples 共Fig. 6兲 show that the crystallization of
`amorphous Ta2N film occurs at about 500 °C, and the forma-
`tion of Cu3Si and TaSi2 after annealing at 700 and 750 °C,
`respectively. The crystallization temperature of amorphous
`Ta2N reported here 共500 °C兲 appears to be a little bit lower
`than that of Sun et al.,27 who reported that a mixture of
`amorphous and crystalline Ta2N film crystallized after an-
`nealing at 600 °C for 65 min. It appears that there is still a
`
`FIG. 6. XRD patterns of the Cu/amorphous Ta2N/Si sample after annealing
`at various temperatures: 共a兲 as-deposited; 共b兲 500 °C; 共c兲 700 °C; and 共d兲
`750 °C.
`
`FIG. 8. AES depth profiles of the Cu/Ta/Si samples: 共a兲 as-deposited; 共b兲
`600 °C, 1 h annealing; and 共c兲 650 °C, 1 h annealing.
`
`J. Vac. Sci. Technol. B, Vol. 14, No. 5, Sep/Oct 1996
`
`Page 5 of 8
`
`

`

`3267
`
`Min, Chun, and Kim: Tantalum and tantalum nitrides
`
`3267
`
`FIG. 9. AES depth profiles of the Cu/amorphous Ta2N/Si samples: 共a兲 as-
`deposited; 共b兲 700, 1 h annealing; and 共c兲 750 °C, 1 h annealing.
`
`FIG. 10. AES depth profiles of the Cu/TaN/Si samples: 共a兲 as-deposited; 共b兲
`650 °C, 1 h annealing; and 共c兲 750 °C, 1 h annealing.
`
`Cu overlayer had little effect on reducing the crystallization
`temperature of amorphous W–Si film, while an Al overlayer
`significantly reduces the crystallization temperature. In order
`to check the effect of Cu on the crystallization temperature of
`amorphous Ta2N phase, a Ta2N film without Cu overlayer
`has been annealed. It is observed that the crystallization of
`amorphous Ta2N occurs at the same temperature, irrespec-
`tive of the presence of a Cu layer. Finally, Fig. 7 shows the
`XRD results of the Cu/TaN/Si film which shows no indica-
`tion of reaction even after annealing at 750 °C.
`Additional results have been obtained in the AES depth
`profiling. In Fig. 8, we note that Cu diffuses deep into the Si
`substrate after annealing at 650 °C. It also shows that exten-
`sive intermixing of Ta and Si has occurred at the surface.
`Figure 9 shows the AES depth profiles of the Cu/amorphous
`Ta2N/Si sample. The barrier failure mode is different com-
`pared to that of the pure Ta 共Fig. 8兲. Figure 9 shows that Cu
`still remain at the surface even though the Ta and Si signal
`significantly intermixed after annealing at 750 °C. This result
`clearly demonstrates that diffusion of Cu into the Si substrate
`
`is significantly slower in the case of a Ta2N layer than it is in
`the case of a Ta. Finally, Fig. 10 shows a series of AES
`depth profiles of the Cu/TaN/Si diffusion couple. As is ex-
`pected from the XRD results, no significant intermixing oc-
`curs even after annealing at 750 °C. The evaluation of barrier
`properties using AES depth profile has two shortcomings in
`our case. One is that the profiles of as-deposited films do not
`show abrupt slope at the interface due to nonuniform etching
`of film during AES depth profile. The other is roughness of
`annealed Cu film surface caused by grain growth during heat
`treatment, which makes the profiles of annealed samples
`worse than that of as-deposited one.
`The Si surface of annealed samples was also examined by
`using optical microscopy after Secco etching 共Fig. 11兲. In
`case of Cu/Ta/Si sample, we first observed the formation of
`etch pits on samples annealed at 500 °C. Both the size and
`the density of etch pits increase with the annealing tempera-
`ture as is shown in Fig. 12. Using cross-sectional TEM, Park
`and Kim13 noted that the initial failure of the Cu/TiN/Si
`sample occurs by the diffusion of Cu into the Si substrate
`
`JVST B - Microelectronics and Nanometer Structures
`
`Page 6 of 8
`
`

`

`3268
`
`Min, Chun, and Kim: Tantalum and tantalum nitrides
`
`3268
`
`FIG. 11. A series of optical micrographs showing the surface of Si after
`Secco etching: 共a兲 450 °C, 1 h annealing; 共b兲 500 °C, 1 h annealing; 共c兲
`550 °C, 1 h annealing; and 共d兲 600 °C, 1 h annealing.
`
`which resulted in the formation of the crystalline defects
`共dislocations兲 decorated with small size of Cu3Si precipi-
`tates. From the previous work, it is concluded that the for-
`mation of etch pits by Secco etching is related to the forma-
`tion of these defects in the Si substrate. It should be noted
`that the formation of reaction product phase共s兲 is first iden-
`tified by XRD at 600 °C while the formation of etch pits is
`first observed at 500 °C. This result indicates that the etch pit
`observation is a more sensitive technique to identify barrier
`failure
`temperatures
`than either XRD or AES. On
`Cu/Ta2N/Si samples, the formation of etch pits is not ob-
`served even after annealing at 650 °C. For samples annealed
`at temperature higher than 650 °C, the formation of etch pits
`is not observed since a significant reaction now has occurred
`at the Ta2N/Si interface 共as is shown from the results of XRD
`and AES兲. From these results, we can conclude that the Ta2N
`barrier is good enough to protect the migration of Cu into the
`Si substrate at least up to 650 °C. The barrier failure, in this
`
`FIG. 13. Optical micrographs of the Si surface after Secco etching: 共a兲
`Cu/Ta2N/Si sample after annealing at 650 °C and 共b兲 Cu/TaN/Si sample
`after annealing at 750 °C.
`
`case, occurs by the interfacial reactions between Ta2N and
`Si. The Cu/TaN/Si sample demonstrates a clean surface even
`after annealing at 700 °C. Etch pits are first observed in the
`samples annealed at 750 °C as is shown in Fig. 13. The in-
`terface between TaN and Si is quite stable even up to 800 °C.
`Due to the strong thermal stability of this interface, the ulti-
`mate failure of the diffusion barrier occurs by the migration
`of Cu into the Si substrate.
`Our results thus indicate that the barrier failure of the
`tantalum and its nitrides occurs by two different mecha-
`nisms. One is the diffusion of Cu into the Si substrate
`through the barrier layer which resulted in the formation of
`crystalline defects and Cu3Si precipitates in the Si substrates.
`This is the predominant failure mechanism for Cu/Ta/Si and
`Cu/TaN/Si samples. The other mechanism is by the chemical
`reactions between the barrier layer and the Si substrate as is
`demonstrated in the Cu/Ta2N/Si samples. According to the
`Cu–Ta–N ternary phase diagram shown in Fig. 14共a兲, Cu is
`to Ta, Ta2N, and
`thermodynamically stable with respect
`TaN. However, the ternary phase diagram drawn from the
`Gibbs free energy data at 900 °K 关Fig. 14共b兲兴 shows that
`both tantalum and its nitrides 共Ta2N and TaN兲 are thermody-
`namically not stable with respect to Si. However, our results
`suggest that the interfacial reaction between TaN and Si oc-
`curs at much higher temperature than that of between Ta2N
`and Si. For this reason, the barrier failure of Ta2N occurs by
`the interfacial reaction while the barrier failure of TaN oc-
`curs by the diffusion of Cu through the barrier layer.
`
`FIG. 12. Etch pit density of the Cu/Ta/Si sample as a function of annealing
`temperature.
`
`FIG. 14. Isothermal section of a ternary phase diagram of 共a兲 Cu–Ta–N and
`共b兲 Ta–N–Si systems drawn at 900 K.
`
`J. Vac. Sci. Technol. B, Vol. 14, No. 5, Sep/Oct 1996
`
`Page 7 of 8
`
`

`

`3269
`
`Min, Chun, and Kim: Tantalum and tantalum nitrides
`
`3269
`
`IV. CONCLUSION
`
`We have investigated the evolution of the microstructure
`and phase共s兲 of the Ta–N systems by varying the nitrogen
`content in the film. It is identified that the as-deposited pure
`Ta film forms a mixture of ␤-Ta and bcc-Ta with a grain size
`of about 20 to 30 nm. When a small amount of nitrogen is
`incorporated in the sputtering gas, the phase changes from a
`mixture of ␤-Ta and bcc-Ta to pure bcc-Ta without a change
`in grain size. It is also identified that both amorphous and a
`mixture of amorphous and crystalline Ta2N phase can be
`formed when the nitrogen content in the film is around 33
`at. % and the crystalline fcc-TaN phase is formed when the
`nitrogen content in the film is from about 40 to 48 at. %.
`By using sheet resistance measurements, XRD, AES
`depth profiles, and etch pit observations after Secco etching,
`it is identified that the diffusion barrier property is signifi-
`cantly enhanced as the nitrogen content in the film is in-
`creased. Importantly, we note that there are two different
`mechanisms of barrier failure; one is by the migration of Cu
`into the Si substrate by diffusion through the barrier layer
`and the other is the interfacial reactions between the barrier
`layer and the Si substrate. Ta apparently fails initially by the
`diffusion of Cu into the Si substrate at 500 °C and by the
`formation of CuSi3 and TaSi2 at higher temperatures. On the
`other hand, for the case of Ta2N film, the barrier failure
`occurs at around 700 °C by the chemical reaction between
`Ta2N and Si. Finally, the TaN film, deposited as an fcc crys-
`talline phase, is identified as a stable barrier up to 700 °C and
`the barrier failure occurs by the migration of Cu into the Si
`substrate.
`
`ACKNOWLEDGMENTS
`
`This research was funded in part by the Ministry of Edu-
`cation through the Inter-University Semiconductor Research
`Center at Seoul National University 共IRS-94-E-1017兲 and in
`part by the Ministry of Science and Technology of Korea
`
`through Electrical Telecommunication Research Institute
`共ETRI兲. The authors are also grateful to Dr. Moshe Eizen-
`berg and Dr. Ivo Raaijmakers at Applied Materials, Inc. for
`the critical review of this manuscript.
`
`1T. Nitta, T. Ohmi, T. Hoshi, S. Sakai, K. Sakaibara, S. Imai, and T.
`Shibata, J. Electrochem. Soc. 140, 1131 共1993兲.
`2J. Tao and N. W. Cheung, IEEE Electron Device Lett. 14, 249 共1993兲.
`3M. O. Abelfotoh and B. G. Stevensson, Phy. Rev. 44, 12 742 共1991兲.
`4A. Broniauowski, Phys. Rev. Lett. 62, 3074 共1989兲.
`5K. Holloway, P. M. Fryer, C. Cabral, Jr., J. M. E. Harper, and P. J.
`Bailey, J. Appl. Phys. 71, 5433 共1992兲.
`6P. Catania, J. P. Doyle, and J. J. Cuomo, J. Vac. Sci. Technol. A 10, 3318
`共1992兲.
`7C. A. Chang, J. Appl. Phys. 67, 7348 共1990兲.
`8E. Kolawa, J. S. Chen, J. S. Reid, P. J. Pokela, and M.-A. Nicolet, J.
`Appl. Phys. 70, 1369 共1991兲.
`9C. A. Chang, J. Appl. Phys. 67, 6184 共1990兲.
`10S.-Q. Wang, S. Suthar, C. Hoeflich, and B. J. Burrow, J. Appl. Phys. 73,
`2301 共1993兲.
`11J. O. Olowolafe, J. Li, and J. W. Mayer, J. Appl. Phys. 68, 6207 共1990兲.
`12S.-Q. Wang, I. J. M. M. Raaijmakers, B. J. Burrow, S. Suthar, S. Redkar,
`and K. B. Kim, J. Appl. Phys. 68, 5176 共1990兲.
`13K. C. Park and K. B. Kim, J. Electrochem. Soc. 共in press兲.
`14A. Charai, H. E. Hornstrom, O. Thomas, P. M. Fryer, and J. M. E. Harper,
`J. Vac. Sci. Technol. A 7, 784 共1989兲.
`15R. E. Thomas, K. J. Guo, D. B. Aaron, E. A. Dobisz, J. H. Perepezko, and
`J. D. Wiley, Thin Solid Films 150, 245 共1987兲.
`16S.-Q. Wang, MRS Bull. Aug. 共1994兲.
`17Binary Phase Diagram, edited by T. B. Massalski 共The Materials Infor-
`mation Society, Materials Park, 1990兲.
`18B. Mehrotra and J. Stimmell, J. Vac. Sci. Technol. B 5, 1736 共1987兲.
`19S. Wolf and R. N. Tauber, Silicon Processing for the VLSI Era 共Lattice
`Press, CA, 1987兲, Vol. 1, p. 533.
`20P. Catania, R. A. Roy, and J. J. Cuomo, J. Appl. Phys. 74, 1008 共1993兲.
`21L. A. Clevenger, A. Mutscheller, J. M. E. Harper, C. Cabral, Jr., and K.
`Barmak, J. Appl. Phys. 72, 4918 共1992兲.
`22P. N. Baker, Thin Solid Films 14, 3 共1972兲.
`23L. G. Feinstein and F. C. Livermore, Thin Solid Films 16, 129 共1973兲.
`24A. Noya, K. Sasaki, and M. Takeyama, Jpn. J. Appl. Phys. 32, 911
`共1993兲.
`25M. H. Rottersman and M. J. Bill, Thin Solid Films 61, 281 共1979兲.
`26J. S. Reid, E. Kolawa, R. P. Ruiz, and M.-A. Nicolet, Thin Solid Films
`236, 319 共1993兲.
`27X. Sun, E. Kolawa, J. S. Chen, J. S. Reid, and M.-A. Nicolet, Thin Solid
`Films 236, 347 共1993兲.
`28R. E. Thomas, J. H. Perepezko, and J. D. Wiley, Appl. Surf. Sci. 26, 534
`共1986兲.
`29I. Barin, Thermochemical Data of Pure Substances 共VCH, New York,
`1989兲.
`
`JVST B - Microelectronics and Nanometer Structures
`
`Page 8 of 8
`
`