Oxidation and removal mechanisms during chemical–mechanical planarization

Wear 263 (2007) 1477–1483

Oxidation and removal mechanisms during chemical–mechanical planarization D. Ng a , M. Kulkarni a , J. Johnson b , A. Zinovev b , D. Yang c , H. Liang a,∗ a

b

Mechanical Engineering, Texas A&M University, College Station, TX 77843-3123, USA Energy Technology Division, Argonne National Laboratory, 9700 South Cass Ave., Argonne, IL 60439, USA c Hysitron Inc., 10025 Valley View Rd., Minneapolis, MN 55344, USA Received 15 August 2006; received in revised form 3 November 2006; accepted 30 November 2006 Available online 23 May 2007

Abstract This paper studies surface properties of metals during chemical–mechanical polishing (CMP). In order to pinpoint the effects of chemistry and mechanical impacts separately, during CMP, we polished Cu and Al surfaces using two distinct slurries of hydrogen peroxide (H2 O2 ) and alumina nanoparticles. After polishing, detailed characterization of the surface quality and chemical composition were conducted using scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), and nanoindentation techniques. It was found that nanoparticles were effective in removing surface materials while passivation provides a high quality layer. The application of H2 O2 slurries in combination with friction stimulation produced an oxide layer. Depending on the nature of metals, it was found that Cu forms an active and non-equilibrium oxide layer while Al has a stable one. The oxide layer resulted from two competing mechanisms, passivation and abrasion. It was deduced from kinetics that oxidation dominated the initial formation of the surface layer while mechanical sweeping determined the final film thickness. New material-removal mechanisms are proposed. © 2007 Elsevier B.V. All rights reserved. Keywords: Chemical–mechanical planarization (CMP); Passivation; Abrasion; Friction; Wear

1. Introduction Mechanical polishing plays an important role in metal fabrication [1]. The polishing mechanisms involve friction that activates surface reaction sites by bond stretching, bond breaking, reformation, mass transport of ions, and nucleation [2–12], as well as increased diffusivity [13], improved boundary lubrication of steel [14–16], metals generating metallic bonds, plastic deformation, mechanically mixed layers during dry sliding, and friction-stirred welding [17]. CMP has been used in the microelectronic industry as a major planarization process step in making chips since the 1980s [18–21]. It generates a super smooth surface with an average ˚ across a 300 mm wafer. Metal roughness of less than 10 A CMP is possible when a passivation layer is formed and then removed through nanoabrasive particles. Oxide CMP involves interaction between abrasive silica particles and the oxide sur-

∗

Corresponding author. E-mail address: [email protected] (H. Liang).

0043-1648/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2006.11.023

face, where material is removed through a “snow-ball” action [22]. It has been reported that oxide layers of silicon, tungsten, and copper are formed after CMP [23–35]. These metal-oxide layers are generally less than a few nanometers in thickness on supersmooth wafer surfaces (∼250 mm in diameter). Metals have high surface energies and oxides have lower ones. The surface energy determines whether a material wets another material and forms a uniform adherent layer. Such a uniformly adherent layer might benefit from friction as it enhances diffusion of surface atoms and reduces residual stress, chemical reactions, and any misfit between film and bulk. Our recent XPS work on copper CMP has shown that nonequilibrium CuO and Cu(OH)2 are formed during polishing [36,37]. This means that mechanical stimulation during CMP plays a crucial role. In order to understand the synergy of the chemical–mechanical process here, we will study the chemical and mechanical effects separately. This allows us to simplify our study and identify those reactions in two body wear. The focus will be on the properties of surfaces in non-equilibrium states. The approach used here is to conduct CMP experiments that

1478

D. Ng et al. / Wear 263 (2007) 1477–1483

Table 1 Physical and mechanical properties of metal materials

Copper Aluminum

Crystal structure

Density (g/cm3 )

Tensile modulus (GPa)

Yield strength (MPa)

fcc fcc

8.96 2.70

129.8 70.6

270 110–170

pinpoint mechanical removal verses oxidation. This is further enhanced through surface characterize using XPS, SEM, and nanoindentation techniques. 2. Experimental procedure 2.1. Materials Two metals, copper and aluminum, were investigated in this study. Samples were cut from copper and aluminum wafers. The physical and mechanical properties are shown in Table 1. The polished area of these samples measured 10 mm × 10 mm. Polyurethane pads (Rohm & Haas IC1000) were used as counterparts to rub against the metal surfaces. Three types of polishing slurries were prepared; they are pure deionized water, water containing ␥-alumina (Buehler), and a slurry containing H2 O2 . Their composition and pH are listed in Table 2. As mentioned earlier, these three slurries were prepared in order to precipitate the desired interaction. 2.2. Polishing experiments Polishing experiments were conducted with a disk-on-disk setup and reciprocating motion (CETR’s UMT). Frictional motion is produced by a scotch yoke mechanism driven by a variable speed motor. The upper sample was loaded against the lower sample by a 2D force sensor, which enabled the measurement of the sliding friction force as well as controlling the loading force. The testing conditions are as follows: normal load at 5 N, reciprocating speed at 200, 400, and 800 rpm, sliding time at 30 min, and operating under ambient conditions. Prior to each test, the metal samples were cleaned using acetone. During experiments, the slurry was applied to cover the pad surface completely. The friction coefficients were recorded and tests were repeated at least three times under each condition. Before polishing, average surface roughness and wetting angle were measured. The surface roughness was measured using a profilometer (Talysurf 3+) with a stylus tip of radius 5 ␮m. Data shown in the following sections are the average of four roughness values. The contact angle measurement was carried out using a goniometer, the water droplet being delivered

onto the sample surface using a syringe. The measurement was made after waiting for 20 s when the droplet is stabilized. The contact angle data was also the average of four measurements, made at different locations on the metal surface. After polishing, sample surfaces were characterized using ellipsometry, scanning electron microscopy (SEM), X-ray photoemission spectroscopy (XPS), and nanoindentation. XPS was used to study surface chemical compositions of the films. XPS data was collected using Mg K␣ (1253.6 eV) radiation with HEA in FAT mode (in vacuum of 3 × 10−10 Torr). Spectra calibration was carried out using a Gold XPS line Au4f7/2 (BE 84 eV). All data were corrected for inelastic scattering by subtracting a Shirley background from the raw spectra. Peak fits were done using pseudo-Voigt shape peaks with different relative content of Gaussian and Lorentzian components. The Ar+ ion beam was used for depth profiling of the metal surface. The ion beam current was 30 ␮A/cm2 and a voltage of 3 kV. Average sputtering rate, which was calculated based on Sigmung’s theory was 4 and 3.7 nm/min for copper and aluminum, respectively [38]. Nanoindentation was performed on four polished samples using a Hysitron’s TriboIndenter® . The samples were labeled as Copper wafer No. 1 polished by 3 wt%. H2 O2 (Cu 1); Copper wafer no. 2 polished by 3 wt.% Al2 O3 (Cu 2); Aluminum wafer no. 1 polished by 3 wt.% H2 O2 (Al 1); and Aluminum wafer no. 2 polished by 3 wt.% Al2 O3 (Al 2). Eight or more indentations were performed on each sample using a 142.3◦ Berkovich diamond indenter tip with a triangular load function consisting of a 5 s loading segment and a 5 s unloading segment. A minimum normal load of 1000 ␮N and maximum normal load of 11,000 ␮N were applied on all four samples for the tests. A high normal load was applied to get reproducible data. In addition, in situ surface probe microscope (SPM) images were obtained from all surfaces to estimate sample roughness. 3. Results and discussions 3.1. Friction and nanomechanical properties The average friction coefficients were obtained from four tests and results are shown in Fig. 1. The shaded bars are friction coefficient and the top portion shows the standard deviation. It is seen that from the slurries tested, both copper and aluminum have friction coefficients between 0.4 and 0.6. The exception is that of Al polished with a slurry containing alumina. This behavior is explained later. Nanoindentation tests were performed to obtain the reduced modulus (Er ) and the hardness (H). Four samples were evaluated—Cu 1, Cu 2, Al 1 and Al 2. Prior to the nanoin-

Table 2 The composition and pH of slurries 3 wt.% ␥-alumina

DI water pH 5 Copper Aluminum

√

pH 7.5 √ √

pH 10 √

pH 5 √

pH 7.5 √ √

3 wt.% H2 O2 pH 10 √

pH 5 √

pH 7.5 √ √

pH 10 √

D. Ng et al. / Wear 263 (2007) 1477–1483

1479

Fig. 3. Multiple load–displacement plots of Al 1 (pink) and Al 2 (blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)

Fig. 1. Average friction coefficient of metals CMP. (a) Cu and (b) Al.

dentation tests, in situ atomic force microscope (AFM) was used to estimate the surface roughness of these samples. Cu 1 and 2 showed the root-mean-square (RMS) roughness values of ∼20 and ∼2 nm, respectively. Al 1 and 2 showed the RMS roughness values of ∼28 and ∼15 nm, respectively. In order to obtain reproducable Er and H data, high normal loads (with large indentation depths) are used for nanoindentation. Results are shown in Figs. 2 (Cu samples) and 3 (Al samples). The figures show that the two copper surfaces are visibly different in depth under the same load, while the depth for Al 2 is greater than for Al 1. Table 3 summarizes the nanoindentation results. The reduced modulus and hardness data show a slight difference in material properties between the samples polished by H2 O2 and by Al2 O3 ; the polished copper surface seems to have significantly higher values than that of aluminum. The Vickers hardness of pure copper is between 49 and 87 and pure aluminum is between 21 and 48 [39]. The high hardness is shown in the nanoindentation data for Cu (Table 3). At this stage, we do not know the actual hardness of the metal oxides

due to the very thin layer. However, we should note that the friction value for copper is lower than that of aluminum. This may be due to the fact that the hard copper surface is less deformed than the soft aluminum so that sliding is relatively favorable. 3.2. Surface properties Ellipsometry data for Cu and Al are shown in Fig. 4(a and b). The delta gives a measure of the change in oxide film thickness of the metal. The figures show that when the polishing surface speed changes (rpm value as the x-axis), delta changes slightly for water and Al2 O3 slurries. However, delta changes more when Cu and Al were polished in H2 O2 slurries. This indicates the strong effect of H2 O2 on surface passivation. The contact angle and average surface roughness versus speed (rpm) are shown in Table 4. Contact angle changed when H2 O2 was used as a polishing slurry but not when other slurries were Table 3 Summary of results from nanoindentation Load (␮N) Copper 1 1200 4000 6000 10000 Aluminum 1 1200 3000 6000

Fig. 2. Multiple load–displacement plots of Cu 1 (light blue) and Cu 2 (pink). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)

Reduced modulus (GPa) 169 159 160 163

± ± ± ±

55 3 17 5

76 ± 6 75 ± 3 63 ± 3

Copper 2 1200 4000 6000 10000

163 150 160 167

Aluminum 2 1200 3000 6000

61 ± 5 64 ± 1 73 ± 8

± ± ± ±

18 8 2 2

Hardness (GPa) 3.0 3.0 2.9 3.2

± ± ± ±

1.0 0.2 0.5 0.3

0.9 ± 0.3 0.8 ± 0.1 0.7 ± 0.1 3.0 2.9 2.9 3.2

± ± ± ±

0.1 0.0 0.2 0.0

0.8 ± 0.1 0.7 ± 0.0 0.7 ± 0.1

Contact depth (nm) 111.6 227.7 280.9 355.1

± ± ± ±

29.3 11.1 25.2 16.5

210.8 ± 37.6 379.3 ± 35.7 597.2 ± 33.1 105.2 223.3 283.1 356.4

± ± ± ±

3.5 1.0 5.1 2.4

231.8 ± 21.0 422.5 ± 10.0 672.1 ± 24.0

1480

D. Ng et al. / Wear 263 (2007) 1477–1483

used. Most roughness values were low and did not vary much with polishing speed with the exception of the low polishing speed of 200 rpm, where H2 O2 for neutral pH is shown to give a rough surface. SEM micrographs are shown in Fig. 5, where (a and b) are for Al surfaces and (c and d) for Cu surfaces. Fig. 5(a) shows pits due to corrosion on the surface while (b) shows more scratches due to mechanical abrasion. The most interesting one is Fig. 5(c) that shows a passivation layer of H2 O2 on the copper surface. The other micrographs exhibit regularly polished or scratched surfaces as already stated. To study the chemical nature of these films, we decided to conduct XPS experiments.

3.3. XPS analysis

Fig. 4. Surface properties of metals measured with an ellipsometry.

Table 4 Contact angle and surface roughness as a function of polishing speed Speed (rpm)

DI water pH 7.5

(a) Cu Contact angle 0 82.63 200 75.8 400 71.37 800 73 Surface roughness (␮m) 0 0.13 200 0.083 400 0.09 800 0.103 (B) Al Contact angle 0 84.17 200 70 400 55.83 800 59.67 Surface roughness (␮m) 0 0.37 200 0.66 400 0.38 800 0.42

H2 O2

Al2 O3

pH 10

pH 7.5

pH 10

pH 7.5

pH 10

82.63 76.53 70.07 74.6

82.63 34.57 51.07 15.13

82.63 48.73 26.63 23.13

82.63 72.53 81.3 82.2

82.63 64.5 67.83 70.63

0.13 0.07 0.063 0.057

0.13 0.37 0.073 0.137

0.13 0.22 0.107 0.083

0.13 0.153 0.073 0.1

0.13 0.07 0.08 0.053

84.17 78.73 65.6 71.4

84.17 37.93 37.63 27.7

84.17 32.93 35.97 24.33

84.17 56 46.17 56.13

84.17 23.6 30.63 19.9

0.37 0.41 0.34 0.37

0.37 0.49 0.36 0.35

0.37 0.67 0.42 0.53

0.37 0.81 0.54 0.68

0.37 0.76 0.65 0.52

A “survey” spectrum was taken over a binding energy ranging from 0 to about 1000 eV for the aforementioned samples. After the elements contained in the film were identified, a detailed series of spectra were taken over the specific range of the elements of interest. The XPS analysis results are shown in Fig. 6. Cu 1 and Cu 2 spectra are shown in Fig. 6(a), whereas Al 1 and Al 2 are in (b). The Cu 1, which was polished in H2 O2 , had a mosaic surface according to SEM. Before any treatment, Cu 1 was found to contain Cu, K, C, and O. This confirms the EDAX analysis for this sample. The sample was then sputtered and analyzed continuously until pure Cu was attained. The total thickness of the surface layer was around 20 nm. Cu 2 was polished in Al2 O3 and showed no film formed in an SEM image. Potassium was not detected in this sample as concurs with the EDAX results. Most peaks shown are similar, although there were some changes between 230 and 300 eV where C and K were found on the Cu 2. The C was from the polishing pad as reported in our previous study [25,26] and K was used as a dispersion addition in the polishing slurry. The counts obtained depend on the depth of the photoelectrons so that the exact concentration of the surface elements is approximate. The XPS results showed no difference between surfaces of Al 1 and Al 2, as shown in Fig. 6(b). This indicates that no matter what slurry is used, the oxidation and removal states of both surfaces are the same. In other words, the tribochemistry during Al-CMP is not affected by the addition of H2 O2 . Comparing Cu and Al, it is interesting to see their oxides have a different reaction to mechanical removal. The Cu 2 does not show any remaining oxide layer after mechanical polishing while Al 1 and 2 show no difference. Apparently, this depends on the bonding strength of the oxide. The Al2 O3 is known for its high hardness and thus it is difficult to remove the oxide through pad sweeping. On the contrary, the CuO and Cu(OH)2 are much softer than Al2 O3 and we have found they are in non-equilibrium state [40,41]. A higher friction coefficient for Al than for Cu in Al2 O3 is seen in Fig. 1. Results indicate that these oxides, in equilibrium or not, dominate the polishing performance. The oxidation is due to tribochemical reactions.

D. Ng et al. / Wear 263 (2007) 1477–1483

1481

Fig. 5. SEM micrographs of metals after CMP.

3.4. Kinetics of film formation Our results showed evidence of a 20 nm thick layer formed on the Cu surface when H2 O2 was present. The formation of such a film was due to passivation as well as mechanical abrasion. Apparently, there is a balance between these two processes. For passivation, the thickness of oxide growth, hox , depends on the time of oxidation, t, through the parabolic relation [42] hox = Kt 0.5

(1)

where K is the parabolic oxidation rate constant. For our analysis, an empirical formula was chosen in the form of the power-law dependence, hox (t) = Kt n + h0

(2)

where hox (t) is the thickness of the newly formed oxide film, n is the polynomial constant (n < 1), and h0 is the initial thickness. For abrasion, the wear volume can be written as [43], V = kw

Fig. 6. XPS spectra for Cu (a) and Al (b).

LS H

(3)

where V is the volume removed, kw the wear coefficient, S the sliding distance, L the load on the sample, and H is the hardness of the material. As mentioned earlier, this research focuses on a simplified polishing process so that we will focus on the two-body abrasive

1482

D. Ng et al. / Wear 263 (2007) 1477–1483

wear only. The two-body-volume wear rate can be expressed as, dV kw L = v dt H

(4)

where v is the velocity of two surface in contact (=dS/dt). If it is assumed that the contact area does not change during sliding contact, the above equation can be written as, dh kw = pv dt H

(5)

Integrating the above equation, we get an expression for the initial thickness, h0 h0 =

kw pvt H

(6)

where p is the pressure applied during polishing (=L/A) and t is the polishing time. Thus the Arrhenius equation for the thickness due to the mechanical (pad) sweeping should be equal to [44]: hwear = h0 e−t/τ ,

where τ =

LP Er

Fig. 7. Illustration of surface film formation under manipulation.

detailed quantitative comparisons will be made using an industrial polisher. 4. Conclusions

(7)

where τ is the exponential decay time constant, L the pad thickness, E the pad modulus, and r is the removal rate. The balance of thickness between oxidation and pad sweeping is,   1 kw h= (8) Kt n + h0 + pvt e−t/τ 2 H As shown in Eq. (1), the oxidation is a function of time exponentially. During CMP, the oxidation takes place in the H2 O2 slurry. Different metals have different constant, K. The subsequent removal of such an oxide layer depends on the bonding strength and stability. Our results showed that the Cu forms an oxide layer that is more active than that of Al. The evidence is seen in the remaining thickness of oxide on Cu 2. Furthermore, the copper oxide is in a non-equilibrium state. How does the resulting oxide layer affect removal? The wear through the pad abrasion is expressed in Eq. (4). This is for a two body abrasive wear model so that we focus on the mechanical abrasion through pad sweeping. Through a few simple steps in Eqs. (5)–(7), we derive an equation that summarizes the resulting thickness of the oxide layer, shown in Eq. (8). It indicates that the resulting oxide thickness in CMP, h, depends on several factors: oxidation constant, oxidation time, initial thickness, wear coefficient, hardness of material, polishing pressure, polishing speed, polishing time, pad thickness, pad modulus and removal rate. Since this work focuses on identifying separate mechanical and chemical reactions in two body wear, a detailed study will be carried in the future. Fig. 7 illustrates competing mechanisms between oxidation and abrasion. The initial thickness of the surface layer increases due to the faster oxidizing rate. After that, the thickness stabilizes, reaching equilibrium. According to Eq. (8), the shape of the balancing thickness curve depends largely on the mechanical abrasion factor, this is only a qualitive illustration. In future,

In this research, we used a non-conventional approach to study CMP mechanisms. Our experiments were able to separate the effects of oxidation and mechanical removal; we focused on the surface properties and their changes that take place during CMP instead of removal rate. The results showed that there was a surface layer formed on copper with H2 O2 . The oxide layer thickness for Cu was 20 nm after CMP, which has not been seen before. Without H2 O2 , however, the polished surface showed either scratches or corroded pits (for aluminum). In addition, sliding friction introduces a nonequilibrium oxide for Cu. This tribo-oxidation was not found in Al. Kinetic analysis of film formation was conducted to balance two competing mechanisms. An equation expressing oxide thickness showed that there are several factors affecting the final thickness: oxidation constant, oxidation time, initial thickness, wear coefficient, hardness of material, polishing pressure, polishing speed, polishing time, pad thickness, pad modulus, and removal rate. The surface residuals and their resulting properties not only reflect the CMP quality, but also reveal fundamental mechanisms of polishing. The mechanical impact and chemical interaction, i.e., manipulation, apparently plays an important role in CMP. This research was able to quantify these effects through kinetic calculations predicting the resulting film thickness. This opens areas of investigation for future research. Acknowledgements This research was in part sponsored by the National Science Foundation (CMS-00239136), Texas Engineering and Experiment Station (TEES), and Texas A&M University. J. Johnson and A. Zinovev were supported by the U.S. Department of Energy, Office of Science, under Contract no. W-31-109-ENG38. Data collection assisted by Helen Xu and Melloy Baker was greatly appreciated.

D. Ng et al. / Wear 263 (2007) 1477–1483

References [1] E. Rabinowicz, Polishing, Sci. Am. 218 (1968) 91–99. [2] F.P. Bowden, D. Tabor, The Friction and Lubrication of Solids, Clarendon Press, Oxford, 1958. [3] T. Ohmi, Trends for future silicon technology, Jpn. J. Appl. Phys. Part 1 33 (1994) 6747–6755. [4] F.W. Preston, The theory and design of plate glass polishing machines, J. Soc. Glass Technol. 11 (1927) 214–256. [5] T.E. Fischer, W.M. Mullins, Chemical aspects of ceramic tribology, J. Phys. Chem. 96 (1992) 5690–5701. [6] H. Tomizawa, T.E. Fischer, Friction and wear of silicon nitride and silicon carbide in water: hydrodynamic lubrication at low sliding speed obtained by tribochemical wear, ASLE Trans. 30 (1986) 41–46. [7] T.E. Fischer, H. Tomizawa, Interaction of tribochemistry and microfracture in the friction and wear of silicon nitride, Wear 105 (1985) 29–45. [8] T.E. Fischer, H. Liang, W.M. Mullins, Tribochemical lubricious oxides on silicon nitride, new directions in tribology, in: L. Pope, L. Fehrenbacher, W. Winer (Eds.), Mater. Res. Soc. Symp. Proc. 140 (1989) 339–344. [9] T. Fischer, Tribochemistry, Ann. Rev. Mater. Sci. 18 (1988) 303–308. [10] G. Heinicke, Tribochemistry, Carl Hanser, Munchen, 1984. [11] J.T. Dickinson, N.S. Park, M.W. Kim, S.C. Langford, A SFM study of a tribochemical process: stress enhanced dissolution, Tribol. Lett. 3 (1997) 69–80. [12] R. Hariadi, S.C. Langford, J.T. Dickinson, Controlling nanometer-scale crystal growth on a model biomaterial with a scanning force microscope, Langmuir 18 (2002) 7773–7776. [13] T. Fischer, M. Sexton, in: P. Lacombe (Ed.), Physical Chemistry of the Solid State: Applications to Metals and Their Compounds, Elsevier, Amsterdam, 1984, pp. 97–106. [14] J.M. Martin, Th. Le Mogne, C. Grossiord, Th. Palermo, Tribochemistry in the analytical UHV tribometer, Tribol. Lett. 3 (1997) 87–94. [15] J.M. Martin, Antiwear mechanisms of zinc dithiophosphate: a chemical hardness approach, Tribol. Lett. 6 (1999) 1–8. [16] K. Varlot, J.M. Martin, C. Grossiord, B. Vacher, K. Inoue, A dual-analysis approach in tribochemistry: application to ZDDP/calcium borate additive interactions, Tribol. Lett. 6 (1999) 181–189. [17] D.A. Rigney, J.E. Hammerberg, Unlubricated sliding behavior of metals, MRS Bull. 23 (1998) 32–36. [18] M.F. Chow, W.L. Guthrie, F.B. Kaufman, US Patent 4,702,792 (1987). [19] J.L. Yeh, G.W. Hills, W.T. Cochran, V.V. Rana, A.M. Garcia, Vacuum 38 (1988) 817. [20] C. Kanta, W. Cote, J. Cronin, K. Holland, P.I. Lee, T. Wright, Submicron wiring technology with tungsten and planarization, in: Proceedings of Fifth International IEEE VLSI Interconnection Conference, 1988, pp. 21– 28. [21] W. Jeffrey, L.D. David, W.L. Guthrie, J. Hopewell, F.B. Kaufman, W.J. Patrick, K.P. Rodbell, R.W. Pasco, A. Nenadic, US Patent 4,954,142 (1990). [22] H. Liang, J. Larsen-Basse, Probable Mechanisms of Si CMP, in: MRS Spring Meeting Presentation, 1999. [23] Y. Li, G.T. Burstein, I.M. Hutchings, Influence of environmental composition and electrochemical potential on the slurry erosion-corrosion of aluminum, Wear 181–183 (1995) 70–79.

1483

[24] Y. Li, G.T. Burstein, I.M. Hutchings, Influence of corrosion on slurry erosion of aluminium, Wear 186–187 (1995) 515–524. [25] Y.L. Wang, J. Wu, C.W. Liu, T.C. Wang, J. Dun, Material characteristics and chemical–mechanical polishing of aluminum alloy thin films, Thin Solid Films 332 (1998) 397–403. [26] H.S. Kuo, W.T. Tsai, Electrochemical behavior of aluminum during chemical mechanical polishing in phosphoric acid based slurry, J. Electrochem. Soc. 147 (2000) 149–154. [27] F.B. Kaufman, D.B. Thompson, R.E. Broadie, M.A. Jaso, W.L. Guthrie, D.J. Pearson, M.B. Small, Chemical–mechanical polishing for fabricating patterned W metal features as chip interconnects, J. Electrochem. Soc. 138 (1991) 3460–3465. [28] E.A. Kneer, C. Raghunath, V. Mathew, S. Raghavan, J.S. Jeon, Electrochemical measurements during the chemical mechanical polishing of tungsten thin film, J. Electrochem. Soc. 144 (1997) 3041–3049. [29] J. Larsen-Basse, H. Liang, Probable role of abrasion in chemo-mechanical polishing of tungsten, Wear 233–235 (1999) 647–654. [30] E. Estragnat, G. Tang, H. Liang, S. Jahanmir, P. Pei, J.M. Martin, Experimental investigation on mechanisms of silicon chemical mechanical polishing, J. Elec. Mater. 33 (2004) 334–339. [31] J.M. Steigerwald, D.J. Duquette, S.P. Murarka, R.J. Gutmann, Electrochemical potential measurements during chemical–mechanical polishing of copper thin films, J. Electrochem. Soc. 142 (1995) 2379–2385. [32] J.M. Steigerwald, S.P. Murarka, R.J. Gutmann, D.J. Duquette, Chemical processes in the chemical mechanical polishing of copper, Mater. Chem. Phys. 41 (1995) 217–228. [33] C.A. Sainio, D.J. Duquette, J. Steigerwald, S.P. Murarka, Electrochemical effects in the chemical–mechanical polishing of copper for integrated circuits, J. Elec. Mater. 25 (1996) 1593–1598. [34] Y. Li, M. Hariharaputhiram, S.V. Babu, Chemical–mechanical polishing of copper and tantalum with silica abrasives, J. Mater. Res. 16 (2001) 1066–1073. [35] G.H. Xu, H. Liang, Effects of electric potential on chemical–mechanical polishing of copper, J. Elec. Mater. 31 (2002) 272–277. [36] M. Kulkarni, D. Greisen, D. Ng, H. Liang, New approaches in investigation of removal mechanisms during copper chemical–mechanical polishing, J. ASTM Int., in press. [37] M. Kulkarni, Tribochemical interaction of microelectronics materials, Ph.D. Thesis, Dept. of Materials Science and Engineering, Texas A&M University, August 2006. [38] P. Sigmund, Theory of sputtering. I. Sputtering yield of amorphous and polycrystalline targets, Phys. Rev. 184 (1969) 383–416. [39] Good Fellow Inc. www.goodfellow.com last accessed on Feb. 1, 2007. [40] H. Liang, J.M. Martin, T.L. Mogne, Interfacial transfer between copper and polyurethane in chemical–mechanical polishing, J. Elec. Mater. 31 (2002) 872–878. [41] H. Liang, J.M. Martin, R. Lee, Influence of oxides on friction during Cu CMP, J. Elec. Mater. 30 (2001) 391–395. [42] M. Rauh, H.U. Finzel, P.Z. Wissmann, Naturforsch 54a (1999) 117. [43] J.F. Archad, Contact and rubbing of flat surfaces, J. Appl. Phys. 24 (1953) 981–985. [44] J. Grillaert, M. Meuris, N. Heylen, K. Devriendt, E. Vrancken, M. Heyns, Proc. CMP-MIC (1999) 79–86.