Cheng-Li Lin, Peng-Sen Chen, and Mao-Chieh Chen
Citation: Journal of Vacuum Science & Technology B 20, 1111 (2002); doi: 10.1116/1.1481863 View online: http://dx.doi.org/10.1116/1.1481863
View Table of Contents: http://scitation.aip.org/content/avs/journal/jvstb/20/3?ver=pdfcov
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing
Articles you may be interested in
Characteristics of copper films deposited on H 2 -plasma-treated TaN substrate by chemical vapor deposition
J. Vac. Sci. Technol. B 20, 1947 (2002); 10.1116/1.1502697
Ionized physical-vapor deposition using a hollow-cathode magnetron source for advanced metallization
J. Vac. Sci. Technol. A 18, 1546 (2000); 10.1116/1.582382
Low temperature plasma-assisted chemical vapor deposition of tantalum nitride from tantalum pentabromide for copper metallization
J. Vac. Sci. Technol. B 17, 182 (1999); 10.1116/1.590533
Structural and electrical properties of chemical vapor deposition tungsten overgrowth on physical vapor deposited and metalorganic chemical vapor deposited TiN adhesion layers
J. Vac. Sci. Technol. B 16, 2013 (1998); 10.1116/1.590122
Chemical vapor deposition of titanium–silicon–nitride films
Ta and TaN for Cu CVD. © 2002 American Vacuum Society. 关DOI: 10.1116/1.1481863兴
I. INTRODUCTION
Copper has been regarded as a potential candidate to re-place Al and its alloys in integrated circuits as the multilevel interconnect material because of its low bulk resistivity共1.67
⍀ cm兲, excellent electromigration resistance, and high
re-sistance to stress-induced voids.1,2To implement copper into metal lines and vias interconnect, the damascene process is developed to cope with the difficult Cu dry-etching problem. However, Cu diffuses rapidly in a Si substrate and forms Cu–Si compounds at low temperatures共about 200 °C兲, caus-ing deep-level traps in Si. Moreover, Cu adheres poorly to dielectric layers and drifts through the oxide under field acceleration.3–5 Therefore, a diffusion/adhesion barrier be-tween Cu and its surrounding layer is necessary for success-ful application of Cu in silicon-based integrated circuits. There are various techniques of copper film deposition, such as chemical vapor deposition 共CVD兲,6,7 conventional and ionized metal plasma共IMP兲, physical vapor deposition,8and electrochemical deposition 共ECD兲, including electroplating and electroless plating;9,10 among these, the CVD method has the advantages of superior step coverage and excellent gap-filling capability for high-aspect-ratio vias and trenches,7 making it a promising technique of Cu film deposition for future integrated-circuit 共IC兲 applications. Although the Cu ECD combined with IMP depositions of thin Cu seed layer and barrier layers provide a suitable solution for IC technolo-gies above 0.25m, the technique of Cu CVD is favorable for future device generations because of the requirement of a more stringent conformal and continuous thin barrier, as well as conformal and void-free Cu film filling into deep subquarter-micron vias.11
The properties of CVD Cu films, such as the growth rate, film texture, surface morphology, impurity contamination, and adhesion to the underlying layer 共barrier兲, are closely related to the initial nucleation of Cu on the substrates.12–15 Variation of nucleation and growth mechanisms of CVD Cu on different substrates leads to different properties of the deposited Cu films. A number of barrier materials have been
used as substrates for chemical vapor deposition of Cu films, such as W, TiW, TiN, Ta, and TaN;6,12,14 –16 among them, TiN, Ta, and TaN exhibit superior barrier properties against Cu diffusion. In this study, we investigate the effect of dif-ferent substrates 共TiN, Ta, and TaN兲 on the nucleation and film properties of CVD Cu using a low-pressure CVD sys-tem built by ourselves for this purpose. Adhesion of the Cu films is also investigated.
II. EXPERIMENT
Figure 1 shows a schematic diagram of the low-pressure CVD system used in this study. The apparatus consists of a reaction chamber 共for Cu film deposition兲, a load-locked chamber共for sample loading/unloading兲, and a direct liquid injection共DLI兲 system for precursor delivery. In the reaction chamber is a shower-head injector through which the Cu precursor is introduced into the reaction chamber in a stream of carrier gas. Under the injector, there is a substrate suscep-tor, which can be heated resistively up to a maximum tem-perature of 400 °C. The susceptor is also rotatable for better uniformity of film deposition. The shower-head injector is movable in vertical direction so that the distance between the injector and the sample can be adjusted. In this study, the susceptor is set to rotate at a speed of 10 rpm, and the injec-tor is set at a position so that it is 2 cm above the suscepinjec-tor. The side wall of the reaction chamber and the precursor in-jector are kept at a temperature of 45 °C by circulation of warm water to prevent Cu deposition of precursor condensa-tion.
The Cu precursor used in this study is Cu 共1,1,1,5,5,5-hexafluoroacetylacetonate兲trimethylvinylsilane 关Cu共hfac兲T-MVS兴 with 2.4 wt % TMVS additive.17,18 The 2.4 wt % TMVS enhances the stability of the precursor.18 The liquid Cu precursor is delivered by the DLI system, which is com-posed of a liquid flow controller 共LFC兲 and a controlled evaporation mixer 共CEM兲. Initially, the liquid precursor is pushed by the N2gas through the LFC. It is then vaporized in
the CEM and mixed with the carrier gas. Helium 共He兲 is used as the carrier gas in this study. The precursor-saturated
carrier gas is introduced into the reaction chamber through the gas injector for the CVD of Cu films.
In this study, TiN, Ta, and TaN were used as the substrates for CVD of Cu films. All of the substrate layers were 50 nm thick, and were sputter-deposited on Si wafers of 4-in.-diam covered with a thermal oxide共500 nm thickness兲. A dc mag-netron sputtering system with a base pressure of 1.5⫻10⫺6 Torr was used to reactively sputter-deposit the TiN, Ta, and TaN substrate layers at a pressure of 7.6 mTorr. The TiN substrate was sputter-deposited using a Ti target共99.99% pu-rity兲 in an Ar/N2 gas mixture; the flow rates of Ar and N2
were 60 and 1.5 sccm, respectively. The Ta and TaN sub-strates were sputter-deposited using a Ta target 共99.99% pu-rity兲 in a pure Ar gas and an Ar/N2gas mixture, respectively;
the flow rate of Ar was 24 sccm for the Ta film deposition, and the flow rates of Ar and N2were 24 and 6 sccm,
respec-tively, for the TaN film deposition.
The substrate wafer 共coated with TiN, Ta, or TaN兲 was loaded into the Cu CVD system. When the pressure of the load-locked chamber reached 10⫺6Torr, the substrate wafer
共together with the substrate holder兲 was transferred to the
reaction chamber for Cu film deposition. Prior to starting the Cu film deposition, the substrate sample was heated to the desired deposition temperature with the He carrier gas flow-ing at 25 sccm and the chamber pressure maintained at 150 mTorr. About 1 h was typically required for the substrate sample to reach the preset temperature. In this study, Cu CVD was performed over a temperature range of 120 to 240 °C at a pressure of 150 mTorr with a precursor flow rate of 0.4 mL/min and a He carrier gas flow rate of 25 sccm. At the end of Cu film deposition, the sample was cooled in ambient He at a pressure of 150 mTorr.
The thickness of Cu films was measured by a DekTek profiler on the patterned Cu films and was verified by cross-sectional scanning electron microscope 共SEM兲 examination. SEM was also used to observe the surface morphology of the deposited Cu films. A four-point probe was employed to measure the sheet resistance. Secondary-ion mass spectrom-etry共SIMS兲 was used to analyze the impurity content in the Cu films. Crystal structure was identified by x-ray diffraction
共XRD兲 analysis. The surface roughness of Cu and various
substrate layers was evaluated by atomic force microscopy
共AFM兲. Rutherford backscattering spectroscopy 共RBS兲 was
used to determine the composition of the substrate layers. The Scotch tape pulling test was used to assess the adhesion between the CVD Cu films and the substrate layers.
III. RESULTS AND DISCUSSION
A. Film properties of CVD Cu films on various substrates
The reactively sputtered TiN, Ta, and TaN substrates have resistivities of 8.50, 0.19, and 0.60 m⍀ cm, respectively, and the compositions of the metal nitrides are TiN1.1and TaN1.2,
as determined by RBS. Figure 2 shows the effective deposi-tion rate of Cu films on different substrates as a funcdeposi-tion of substrate temperature 共Arrhenius plot兲 at a deposition pres-sure of 150 mTorr with a He carrier gas flow rate of 25 sccm and a liquid Cu precursor flow rate of 0.4 mL/min. The ef-fective deposition rate of a Cu film was calculated using the thickness deposited in a period of 10 min. The chemical reaction of Cu CVD using Cu共hfac兲TMVS as a precursor proceeds on the substrate surface by a facile disproportion-ation as follows:19
2Cu⫹1共hfac兲TMVS共g兲→2Cu⫹1共hfac兲TMVS共s兲, 共1兲 2Cu⫹1共hfac兲TMVS共s兲→2Cu⫹1共hfac兲共s兲⫹2TMVS共g兲, 共2兲 2Cu⫹1共hfac兲共s兲→Cu共s兲⫹Cu⫹2共hfac兲2共s兲, 共3兲
Cu⫹2共hfac兲2共s兲→Cu⫹2共hfac兲2共g兲, 共4兲 where共g兲 denotes ‘‘gas phase’’ and 共s兲 denotes ‘‘adsorbed on substrate surface.’’ The reaction step 3 关Eq. 共3兲兴 is the key step of Cu nucleation on the substrate surface, which in-volves a process of electron exchange between the adsorbed Cu⫹1共hfac兲 and the substrate surface. Thus, it is easier to deposit Cu films on a conducting substrate than on an insu-lating substrate. Because the chemical reaction of Cu CVD FIG. 1. Schematic diagram of low-pressure Cu CVD apparatus used in this
study.
FIG. 2. Deposition rate of Cu film vs substrate temperature共Arrhenius plot兲 at a pressure of 150 mTorr.
involves a thermal dissociation of the Cu precursor关Eq. 共2兲兴, a higher temperature would result in a higher rate of deposi-tion. The activation energy Ea was determined to be 9.04, 10.50, and 7.35 kcal/mole for TiN, Ta, and TaN substrates, respectively, by the Arrhenius equation:
R⫽R0 exp共⫺Ea/kT兲, 共5兲
where R is the deposition rate, R0is known as the Arrhenius
pre-exponential constant or frequency factor, k is the Boltz-mann constant, and T is the absolute temperature. The Cu films deposited on TiN and Ta substrates have a higher ef-fective deposition rate than those deposited on the TaN sub-strate in the surface-reaction-controlled region; however, in the mass-flow-controlled region, Cu films on different sub-strates have nearly the same effective deposition rate. The resistivity of the Cu film was calculated using the measured
the substrate. Among the three substrate layers used in this study, Cu films deposited on TiN substrates revealed the low-est resistivity over the entire temperature range of 120 to 240 °C, and the resistivity of the film deposited at 160 °C was determined to be 1.9 ⍀ cm. Figure 4 illustrates the SIMS depth profiles of Cu films deposited on TiN, Ta, and TaN substrates at a pressure of 150 mTorr. The contamina-tion共impurities兲 of Cu films is associated with the dispropor-tionation of Cu共hfac兲TMVS precursor. It has been reported that TMVS is easily exhausted and not a contributor of con-tamination at high vapor pressures, however, the hfac mol-ecule is the source of observed impurities of fluorine 共F兲, carbon 共C兲, hydrogen 共H兲, and oxygen 共O兲.19,21,22The con-tent of impurities in the Cu film deposited at 220 °C is higher than that in the film deposited at 160 °C关Figs. 4共a兲 and 4共b兲兴; this condition results in higher resistivity for the Cu film deposited at 220 °C共Fig. 3兲. At the same deposition tempera-ture of 160 °C, on the other hand, the content of impurities in the Cu film deposited on TiN substrate is lower than those in the Cu films deposited on the Ta and TaN substrates 关Figs.
FIG. 3. Resistivity of Cu film vs deposition temperature for Cu films depos-ited at a pressure of 150 mTorr.
FIG. 4. SIMS depth profiles of Cu films deposited on共a兲 TiN substrate at 160 °C,共b兲 TiN substrate at 220 °C,
共c兲 Ta substrate at 160 °C, and 共d兲 TaN
substrate at 160 °C. The deposition time is 10 min.
4共a兲, 4共c兲, and 4共d兲兴; this observation of impurity content is also consistent with the measured film resistivity 共Fig. 3兲. Notably, the impurities at the Cu/substrate interface have a higher content and broader distribution for the Cu films de-posited on Ta and TaN substrates compared to that dede-posited on the TiN substrate. Moreover, it is found that the impurities easily accumulated at the substrate surface in the initial stage of CVD Cu film deposition, especially for fluorine impuri-ties, similar to the report in literature.21
Different underlayer substrates would result in different preferred orientation of the deposited Cu films. Figure 5 shows the XRD spectra for Cu films deposited on various substrates. For the Cu film deposited at 160 °C, the intensity peak ratio of Cu共111兲 to Cu共200兲 reflections on the TiN sub-strate was determined to be 3.39, which is higher than the peak ratios observed on the Cu films deposited on Ta and TaN substrates 关Fig. 5共a兲兴. At a higher deposition tempera-ture of 200 °C, all of the peak ratios increased, especially for the Cu film deposited on the TiN substrate关Fig. 5共b兲兴. Figure 6 illustrates the XRD intensity peak ratio of Cu共111兲 to Cu共200兲 reflections vs deposition temperature of Cu films deposited on the three different substrates. Copper films de-posited on the TiN substrate revealed the highest peak ratio among the three substrates investigated in this study.
B. Nucleation and surface morphology
The surface condition of substrates plays an important role in the Cu CVD. Figure 7 shows the AFM images of the as-deposited TiN, Ta, and TaN substrates. The TiN substrate has a rougher surface than the Ta and TaN substrates, whereas the Ta substrate has the smoothest surface. Different substrate surface properties will have different effects on the nucleation processes of the Cu film, which in turn will lead to different properties of final Cu film. Figure 8 shows the SEM micrographs for Cu nucleations on TiN, Ta, and TaN substrates at 160 °C for 1 and 2 min. During the first minute,
FIG. 5. XRD spectra of Cu films deposited on various substrates at 共a兲 160 °C and共b兲 200 °C.
FIG. 6. XRD peak ratio of Cu共111兲 to Cu共200兲 reflections vs deposition temperature of Cu films deposited on various substrates.
FIG. 7. AFM images showing surface morphology of共a兲 TiN, 共b兲 Ta, and 共c兲
TaN substrates. J. Vac. Sci. Technol. B, Vol. 20, No. 3, MayÕJun 2002
the Cu-containing adspecies 关Cu共hfac兲兴 easily nucleated on the TiN substrate, forming dense, small, and uniformly dis-tributed Cu nuclei 关Fig. 8共a兲兴. On the other hand, the Cu-containing adspecies sparsely nucleated on Ta and TaN sub-strates, especially on the latter, forming less dense Cu nuclei of larger size关Figs. 8共b兲 and 8共c兲兴. After CVD for 2 min, Cu nuclei on the TiN substrate coalesced and grew into dense and large Cu grains 关Fig. 8共d兲兴. On the Ta and TaN sub-strates, however, the subsequent Cu-containing adspecies tended to nucleate on the existing Cu nuclei 共grains兲 rather than on the substrate surfaces, thus forming larger but less densely distributed Cu grains 关Figs. 8共e兲 and 8共f兲兴. We pre-sume that the Cu-containing adspecies 关Cu共hfac兲兴 are easily adsorbed by the TiN substrate and proceed with deposition of Cu via disproportionation as compared to that on Ta and TaN substrates, although the resistivity of TiN is larger than those of Ta and TaN, making it unfavorable for the process of electron exchange. However, there is a higher content of im-purities adsorbed on Ta and TaN substrate surfaces at the initial stage of nucleation; this would degrade the conductiv-ity of Ta and TaN substrate surfaces and thus adversely affect the process of electron exchange, resulting in difficult depo-sition of Cu films on Ta and TaN substrates. TiN film was similarly observed to be a good nucleation layer in metal-organic CVD.23 The dense and uniformly distributed Cu
grains on the TiN substrate indicate that TiN substrate has a higher surface energy or lower interfacial energy, thus result-ing in a smaller wettresult-ing angle共contact angle兲 of Cu nuclei on TiN than that on Ta and TaN substrates.24,25 The smaller wetting angle enhanced the Cu film growth in two dimen-sions 共layer growth兲, forming the most stable and 共111兲 closely packed configuration;24thus, the Cu film on the TiN substrate has a higher 共111兲-preferred orientation than those on Ta and TaN substrates.
Figure 9 illustrates the SEM micrographs of Cu films de-posited at 160 and 200 °C for 10 min on TiN, Ta, and TaN substrates. For the films deposited at 160 °C, Cu grains ap-pear to be agglomerate and the surface morphologies show no obvious difference for all Cu films. For the films depos-ited at 200 °C, however, the Cu films on TiN and Ta sub-strates exhibit a compact arrangement and clear boundaries of Cu grains, whereas the Cu film on the TaN substrate re-veals a loose arrangement of grains mixed with different sizes. The surface roughness of Cu films deposited on vari-ous substrates was analyzed by AFM and illustrated in Fig. 10. The average surface roughnesses共rms兲 of the Cu films on TiN, Ta, and TaN substrates were determined to be 49.5, 34.2, and 55.9 nm, respectively. Presumably, the smoothness of the film surface is directly related to the smoothness of the substrate surface.
FIG. 8. SEM micrographs showing Cu nucleation at 160 °C for 1 min deposition on共a兲 TiN, 共b兲 Ta, and 共c兲 TaN substrates, and for 2 min deposition on共d兲 TiN, 共e兲 Ta, and共f兲 TaN substrates.
C. Adhesion measurement
A simple technique—the Scotch tape pulling test—was used to assess the adhesion between the CVD Cu film and the underlayer substrate. We measured the adhesion of CVD Cu films deposited on TiN, Ta, and TaN substrates at a tem-perature ranging from 120 to 240 °C; results of the adhesion test are summarized in Table I. All of the Cu films deposited on the TiN substrate 共at 120 to 240 °C兲 passed the Scotch tape pulling test. For the Cu films deposited on Ta and TaN substrates, the films deposited at and above 160 °C passed the Scotch tape pulling test, whereas the films deposited at temperatures below 160 °C peeled off after the pulling test. At low deposition temperatures, there is less intermixture of Cu and the substrate materials at the interface than at high deposition temperatures.26 Thus, Cu films deposited at low temperatures tend to have an inferior adhesion. Moreover, the surface condition of the substrate and the size, density, and distribution of Cu nuclei on the substrate surface during the initial stage of Cu CVD play important roles for film adhesion. It was reported that film adhesion can be promoted by rougher substrate surfaces because there is a larger sur-face area than a flat sursur-face.26Because the TiN substrate has a rougher surface, as well as denser and smaller Cu nuclei on its surface, during the initial stage of nucleation than Ta and
TaN substrates, the Cu film adhesive area on the TiN sub-strate is effectively enlarged, leading to superior adhesion of Cu films on the TiN substrate. The adhesion can be degraded by impurities 共F, C, and O兲 absorbed on the substrate surface.14,26Because the impurity content at Cu/Ta and Cu/ TaN interfaces are higher than that at the Cu/TiN interface, this might also contribute to the fact that a Cu film on the TiN substrate has superior adhesion.
Copper CVD possesses a superior void-free via/trench filling capability. We believe that low sticking coefficient and high re-emission frequency of Cu-containing adspecies
关Cu-共hfac兲TMVS兴 are essential to obtain good film conformality
and void-free vias/trenches filling capability.27 Because low sticking coefficient and high re-emission frequency can be achieved by decreasing the substrate temperature, Cu CVD at low temperatures is preferred to obtain a void-free filling for high-aspect-ratio vias/trenches. From this viewpoint, a TiN substrate provides a good nucleation layer for Cu CVD. IV. CONCLUSION
Cu CVD on different substrates, including TiN, Ta, and TaN, at a pressure of 150 mTorr and a temperature ranging from 120 to 240 °C was studied with regard to the physical property, nucleation, and adhesion of the deposited Cu films. FIG. 9. SEM micrographs showing surface morphology
of Cu films deposited at 160 °C on共a兲 TiN, 共b兲 Ta, and
共c兲 TaN substrates, and at 200 °C on 共d兲 TiN, 共e兲 Ta, and 共f兲 TaN substrates. The deposition time is 10 min.
The Cu films deposited on TiN substrates have a number of favorable properties over the films deposited on Ta and TaN substrates. These include lower electrical resistivity, lower impurity contamination, and higher 共111兲-preferred orienta-tion. Moreover, CVD of Cu films on TiN substrates has a shorter incubation time and better film adhesion to the un-derlayer substrate. Presumably, the TiN substrate has a higher surface energy than Ta and TaN substrates, and the high surface energy enhances the film growth in two dimen-sions共layer growth兲, resulting in Cu film with a higher 共111兲-preferred orientation. During the nucleation stage of Cu nu-clei, the dense, small, and uniformly distributed Cu nuclei
共grains兲 on the rough TiN substrate surface 共which is rougher
than Ta and TaN substrate兲 effectively enlarge the adhesive
National Nano Device Laboratory for providing an excellent processing environment.
1A. Jain, T. T. Kodas, R. Jairath, and M. J. Hampden-Smith, J. Vac. Sci.
Technol. B 11, 2107共1993兲.
2N. Awaya, H. Inokawa, E. Yamamoto, Y. Okazaki, M. Miyake, Y. Arita,
and T. Kobayashi, IEEE Trans. Electron Devices ED-43, 1206共1996兲.
3Y. Shacham-Diamand, A. Dedhia, D. Hoffstetter, and W. G. Oldham, J.
Electrochem. Soc. 140, 2427共1993兲.
4A. L. S. Loke, C. Ryu, C. P. Yue, J. S. H. Cho, and S. S. Wong, IEEE
Electron Device Lett. EDL-17, 549共1996兲.
5
J. C. Chiou, H. I. Wang, and M. C. Chen, J. Electrochem. Soc. 143, 990
共1996兲.
6R. Kroger, M. Eizenberg, D. Cong, N. Yoshida, L. Y. Chen, S.
Ra-maswami, and D. Carl, J. Electrochem. Soc. 146, 3248共1999兲.
7
C. H. Lee, K. H. Shen, T. K. Ku, C. H. Luo, C. C. Tso, H. W. Chou, and C. Hsia, IEEE International Interconnect Technology Conference共IITC兲, San Francisco, 2000, p. 242.
8B. Chin, P. Ding, B. Sun, T. Chiang, D. Angelo, I. Hashim, Z. Xu, S.
Edelstein, and F. Chen, Solid State Technol. 41, 141共1998兲.
9
R. L. Jackson, E. Broadbent, T. Cacouris, A. Harrus, M. Biberger, E. Patton, and T. Walsh, Solid State Technol. 41, 49共1998兲.
10Y. Shacham-Diamand and S. Lopatin, Microelectron. Eng. 37Õ38, 77
共1997兲.
11
P. Motte, M. Proust, J. Torres, Y. Gobil, Y. Morand, J. Palleau, R. Pantel, and M. Juhel, Microelectron. Eng. 50, 369共2000兲.
12S. Kim, D. J. Choi, K. R. Yoon, K. H. Kim, and S. K. Koh, Thin Solid
Films 311, 218共1997兲.
13Y. Qian, C. Y. Li, L. T. Koh, J. J. Wu, and J. Xie, Proceedings of VLSI
Multilevel Interconnection Conference共VMIC兲, Santa Clara, CA, 1999, p. 558.
14R. Kroger, M. Eizenberg, D. Cong, N. Yoshida, L. Y. Chen, S.
Ra-maswani, and D. Carl, Microelectron. Eng. 50, 375共2000兲.
15
N. I. Cho and Y. Sul, Mater. Sci. Eng. B 72, 184共2000兲.
16
J. C. Chiou, K. C. Juang, and M. C. Chen, J. Electrochem. Soc. 142, 177
共1995兲.
17J. A. T. Norman, B. A. Mutamore, P. N. Dyer, D. A. Roberts, and A. K.
Hochberg, J. Phys. IV 1, C2-271共1991兲.
18
T. Nguyen, L. J. Charneski, and S. T. Hsu, J. Electrochem. Soc. 144, 2829
共1997兲.
19J. A. T. Norman, D. A. Roberts, A. K. Hochberg, P. Smith, G. A. Petersen,
J. E. Parmeter, C. A. Apblett, and T. R. Omstead, Thin Solid Films 262, 46共1995兲.
20
S. S. Yoon, J. S. Min, and J. S. Chun, J. Mater. Sci. 30, 2029共1995兲.
21P. J. Lin and M. C. Chen, Jpn. J. Appl. Phys., Part 1 38, 4863共1999兲. 22S. Voss, S. Gandikota, L. Y. Chen, R. Tao, D. Cong, A. Duboust, N.
Yoshida, and S. Ramaswami, Microelectron. Eng. 50, 501共2000兲.
23
M. E. Gross and V. M. Donnelly, in Advanced Metallization for ULSI
Applications, edited by V. S. Rana and R. V. Joshi共Materials Research
Society, Pittsburgh, 1991兲, p. 355.
24K. Kamoshida and Y. Ito, J. Vac. Sci. Technol. B 15, 961共1997兲. 25
R. J. Stokes and D. F. Evans, Fundamentals of Interfacial Engineering
共Wiley-VCH, New York, 1997兲, p. 59.
26S. Gandikota, S. Voss, R. Tao, A. Duboust, D. Cong, L. Y. Chen, S.
Ramaswami, and D. Carl, Microelectron. Eng. 50, 547共2000兲.
27A. Burke, G. Braeckelmann, D. Manger, E. Eisenbraun, and A. E.
Kaloy-eros, J. Appl. Phys. 82, 4651共1997兲. FIG. 10. AFM images showing surface morphology of Cu films deposited at
160 °C for 10 min on共a兲 TiN, 共b兲 Ta, and 共c兲 TaN substrates.
TABLEI. Results of the Scotch tape pulling test on the adhesion of CVD Cu
films deposited at different temperatures on various substrates.
Substrate
Cu film deposition temperature共°C兲
120 140 160 180 200 220 240
TiN P P P P P P P
Ta F F* P P P P P
TaN F F* P P P P P
P: Cu film passed the Scotch tape pulling test. F: Cu film peeled off after the Scotch tape pulling test. F*: Cu film partially peeled off after the Scotch tape pulling test.
