This content has been downloaded from IOPscience. Please scroll down to see the full text.
Download details:
IP Address: 140.113.38.11
This content was downloaded on 28/04/2014 at 21:38
Please note that terms and conditions apply.
Effect of Electric Potential and Mechanical Force on Copper Electro-Chemical Mechanical
Planarization
View the table of contents for this issue, or go to the journal homepage for more 2012 Jpn. J. Appl. Phys. 51 036504
(http://iopscience.iop.org/1347-4065/51/3R/036504)
Effect of Electric Potential and Mechanical Force
on Copper Electro-Chemical Mechanical Planarization
Sheng-Wen Chen1, Te-Ming Kung1, Chuan-Pu Liu1, Shih-Chieh Chang2, Yi-Lung Cheng3, and Ying-Lang Wang21Department of Material Science and Engineering, National Chen-Kung University, Tainan 701, Taiwan, R.O.C. 2Institute of Lighting and Energy Photonics, National Chiao Tung University, Hsinchu 30050, Taiwan, R.O.C. 3Department of Electrical Engineering, National Chi Nan University, Nantou 545, Taiwan, R.O.C.
Received September 15, 2011; accepted December 30, 2011; published online March 5, 2012
In this study, the dependence of Cu electrochemical mechanical planarization (ECMP) rate on electric potential and mechanical force in electrolyte is investigated using potentiodynamic analysis, electrochemical impedance spectroscopy (EIS), and X-ray photoelectron spectroscopy (XPS). In chemical etching, CMP, electropolishing, and ECMP processes, the Cu removal rate is mainly affected by the interplay between electric potential and mechanical force. An equivalent circuit is built by fitting the EIS results to explain the behavior of Cu dissolution and Cu passive film. The Cu dissolution rate increased with decreasing charge-transfer time-delay. The resistance of the Cu passive film (Rp) is proportional to the intensity
ratio of Cu2O=½CuðOHÞ2þ CuO. # 2012 The Japan Society of Applied Physics
1. Introduction
The planarization of copper (Cu) damascene structures is becoming increasingly challenging due to the ongoing reduction of feature sizes, along with high packing density, use of ultra-low-k dielectrics, and an increasing number of interconnect layers in silicon integrated circuits.1,2) The chemical mechanical planarization (CMP) technique is limited for such applications, with major problems being instability of slurries, scratches generated on surfaces, various metal contaminants, dishing features, post-cleaning difficulties, selectivity problems, and high cost of equipment and maintenance.3–9) The material integration difficulties described above for Cu CMP with porous low-k dielectrics have led researchers to investigate alternative planarization processes such as electro-polishing and electrochemical mechanical planarization (ECMP). ECMP appears to be a viable approach for the efficient planarization of ultra low-k structures and fulfill the strict technical limita-tions.10–14) ECMP combines aspects of two technologies, namely CMP and electropolishing.15) During polishing, a passive film, such as an oxide film, is formed at high electric potentials and is then removed by mechanical wearing.10,16–19) The factors that affect ECMP include characteristics of the solutions,20–22) electric potential, and mechanical force.10,23) The Cu removal rate for ECMP has been demonstrated to be linearly proportional to electric potential.10) However, at the present time, little is known about ECMP in general,10,11,13,15,20–23) especially its surface reactions and dissolution/polishing mechanisms. In order to obtain an appropriate Cu metallization process, the effect of electric potential and mechanical force in the electrolyte for Cu-ECMP processes should be investi-gated. In this paper, the relationship between polishing conditions and the ECMP removal rate is examined in depth and an equivalent circuit is established to explain the dissolution and polishing behaviors of a Cu film in an H3PO4 electrolyte using electrochemical impedance spectroscopy (EIS). The structure of the Cu surface after ECMP is analyzed using X-ray photoelectron spectroscopy (XPS).
2. Experimental Procedure
The experiments were performed at room temperature in a three-electrode cell constructed from a 350-mL cubic cell made of poly(vinyl chloride) (PVC) material, a 7-cm2 blanket wafer as the working electrode, a 7-cm2 platinum plate as the counter electrode, and an Ag/AgCl reference electrode in a 3 mol/L KCl saturated solution. The blanket wafer was prepared by depositing a 30-nm-thick sputtered TaN diffusion barrier followed by a 100-nm-thick Cu seed layer on a 200-nm-thick SiO2/Si(100) substrate. Subse-quently, a Cu film was electroplated to a thickness of 1.1m in a solution consisting of 0.25 mol/L CuSO4
5H2O, 1.0 mol/L sulfuric acid, and 50 ppm chloride ions. The ECMP operating conditions were: electric voltages ranged from 300 to 1800 mV vs Ag/AgCl, down force ranged from 0.07 to 70 psi, and the rotation speed ranged from 50 to 800 rpm. Soft polishing pads (Rodel Politex pad) were used in the ECMP process. The polishing pads were porous to let the electrolyte (85 wt % H3PO4) through, creating effective electrical pathways between the counter electrode, reference electrode, and working electrode. The cross-sections of polished Cu films were imaged using a field emission scanning electron microscope (FESEM; Hitachi S4800) operated at 15 kV. Electrochemical analyses, including in-situ current-time and potentiodynamic tests, were performed with a variable speed rotator [636 rotating disk electrode (RDE)] and PARSTAT 2273 controlled by PowerCORR software (Princeton Applied Research) in a three-electrode cell. EIS measurements were performed at 300, 500, 800, and 1800 mV versus open circuit potential, and conducted by superimposing an AC signal with an amplitude of20 mV in the frequency range of 100 kHz to 1 Hz. The equivalent circuit was established and simulated using ZSimpWin version 3.1 with EIS data. XPS was acquired with a VG Scientific Microlab photoelectron spectrometer at a base pressure below 1 108Torr using Mg K radiation (1253.6 eV). The time needed for each data point in a single scan was 0.1 s, and the data interval was 0.1 eV. The absolute binding energies of various surface species were calibrated using the C 1s line at 285.3 eV. The inelastic background was removed from the spectra using Shirley’s method.24)E-mail address: ylwang@tsmc.com
3. Results and Discussion
A comparison of the etching/removal rates of a Cu film for chemical etching (CE), CMP, electropolishing, and ECMP is shown in Fig. 1. The polishing conditions are listed in Table I. ECMP was decomposed into CE, CMP, and elec-tropolishing to better understand its mechanisms. 85 wt % H3PO4 was used as the oxidant (for CE and CMP) and electrolyte (for electropolishing and ECMP) in the experi-ments. Politex polishing pads with a down force of 0.14 psi were employed in CMP and ECMP. In Fig. 1, ECMP has the highest Cu removal rate. The percentages of the ratios of the Cu removal rate of CE, CMP, and electropolishing to that of ECMP are 0.1, 1.1, and 10.3%, respectively. For static etching, it is difficult to quickly corrode Cu films in a weak acid (H3PO4). The removal rate of ECMP is much higher than the sum of the other polishing methods, implying that the effects in combination have a much larger impact on the total removal rate than they do individually.
Figure 2(a) shows the relationship between current density (Id) and electric voltage against the reference electrode (Ag/AgCl) during ECMP and electropolishing.Id indicates that Cu metal was dissolved; it is produced at the anode surface at the given potential. In Fig. 2(a), the curve for ECMP was obtained at a rotation speed of 200 rpm with a down force of 0.14 psi, whereas the curve for electropolishing was obtained at a rotation speed of 200 rpm without down force. The negative slope ofId for electropolishing implies that the Cu removal rate decreased with increasing electric potential, which is due to the formation of a passive film.25,26)
In contrast,Idincreases with increasing electric potential for ECMP because the passive film is removed by mechanical force, which includes the down force and rotation speed, upon formation. Therefore, the combined effects of the mechanical force and electric potential become more obvious at higher potentials. Figure 2(b) shows the Cu removal rate as a function of electric potential, down force, and rotation speed in a H3PO4 electrolyte in ECMP, with the other two variables kept constant. It was found that the effect of electric potential on the Cu removal rate of ECMP is more significant than that of the mechanical force. This result is due to the Cu dissolution rate being high at high electric potentials.
Figure 2(c) shows the potential dynamic curves obtained for ECMP by scanning the potential (at 20 mV/s) at down forces of 0 to 70 psi. The curve at 0 down force is typical for a passive metal, featuring a dissolution peak at0:03{0:7 V, a passivation plateau at 0:7{1:6 V, and pit generation followed by O2 evolution for high anodic potentials (above 1.6 V).25,26)The current density thus decreases with increas-ing electric voltage in Fig. 2(a). For curves with a nonzero down force, the passivation plateau region disappears. Id increases with increasing down force at potentials above 0.2 V. For example, whenE ¼ 0:3 V vs Ag/AgCl, as shown in the inset of Fig. 2(c), Id slightly increases from 43.7 to 53.8 mA/cm2 as the down force increases from 0.07 to 70 psi. Rotation speed is also factor in the mechanical force. Figure 2(d) shows the potential dynamic curves of the Cu disk electrode for rotation speeds of 50 to 800 rpm under a down force of 0.14 psi in a H3PO4 electrolyte. In Fig. 2(d), Id increases with the rotation speed at potentials above 0.2 V. For example, whenE ¼ 0:3 V vs Ag/AgCl, as shown clearly in the inset of Fig. 2(d), Id slightly increases from 29.4 to 38.0 mA/cm2as the rotation speed increases from 50 to 800 rpm. The increase ofId is consistent with the results shown in Fig. 2(b).
Figures 3(a)–3(c) shows the Nyquist plots of the Cu films as a function of electric potential, down force, and rotation speed and Cu polished in a H3PO4 electrolyte for ECMP, respectively; the other two variables were kept constant. In the Nyquist plots, the y-axis is the imaginary impedance (reactance) and thex-axis is the real impedance (resistance) in the equivalent circuit associated with the ECMP system. All the plots show two semicircles, one on the left at high frequencies (>1000 Hz) one on the right at low frequencies (<0:1 Hz). The possible chemical reactions of a Cu film in a H3PO4 electrolyte are as follows:26–28)
CuðSÞ ! Cuþþ e; ð1Þ CuðSÞ ! Cu2þþ 2e; ð2Þ 2Cuþþ H2O! Cu2Oþ 2Hþ; ð3Þ Cuþþ 2OH! CuðOHÞ2; ð4Þ Cu2Oþ H2O! 2CuO þ 2Hþþ 2e; ð5Þ Cuþ H3PO4þ H2O ! ½CuOHþ½H 2PO4þ 2Hþþ 2e: ð6Þ The high frequency loop is attributed to the charge-transfer resistance and double-layer capacitance from reactions (1) and (2) and the low frequency loop is attributed to the convective and diffusive mass transport of Cu2O, Cu(OH)2, and CuO at the surface from reactions (3)–(6) in accordance with previous reports.27–29) Furthermore, the
CE CMP Electropolishing ECMP 0 400 800 1200 1600 2000
Etch rate/ Removal rate (nm/min)
Polishing methods
Fig. 1. Cu removal rate for chemical etching (CE), chemical mechanical planarization (CMP), electroploishing, and electrochemical mechanical planarization (ECMP).
Table I. Operation conditions of electric potential and mechanical force for chemical etching (CE), chemical mechanical planarization (CMP), electro-ploishing, and electrochemical mechanical planarization (ECMP).
Electric voltage (mV) Down-force (psi) Speed of rotation (rpm) Solution CE 0 0 0 85 wt % H3PO4 CMP 0 0.14 200 85 wt % H3PO4 Electro-polishing 800 0 200 85 wt % H3PO4 ECMP 800 0.14 200 85 wt % H3PO4 S.-W. Chen et al. Jpn. J. Appl. Phys. 51 (2012) 036504
capacitive loop observed in the high frequency range is related to the relaxation of the double layer since its charging–discharging is a very rapid process.30) In brief, there are two steps for the chemical reactions in the Cu ECMP process: (i) Cu metal is dissolved and ionized [e.g., Cuþ, Cu2þ] by charge-transfer reactions with the electrolyte at a given applied electric potential (the first semicircle of the Nyquist plot), (ii) Cu2þ and Cuþ complexes diffuse to the surface of the Cu film and react with O2 and OH species to form a temporary passive film [e.g., Cu2O, CuO, Cu(OH)2], as a diffusion layer, on the Cu surface.
Figure 3(d) shows the simulated equivalent circuit of the Cu ECMP system obtained using ZSimpWin. The best fit was obtained by matching to the result of the EV800DF0.14SR200 case in Fig. 3(a). In Fig. 3(d),Rsis the bulk solution resistance;Cdl is the double-layer capacitance of surface reductions; Rct is the charge-transfer resistance (associated with the double layer); Cp is the passivative-capacitance of the adsorption layer; andRpis the resistance of the passive layer. In this circuit, the RC element of the double layer and the RC element of the passive film are not in series because the passive film is porous.RctandRpare in
series because the dissolution of Cu films occurs before the formation of the passive film on the surface,31) which are characterized by the left and right semicircles in the Nyquist plots of Fig. 3(d), respectively. The best-fitted data to Figs. 3(a)–3(c) for all the circuit components are summa-rized in Table II. CPEp is the impedance in the constant phase element (CPE) of the Cu passive layer. The values of the system’s capacitances can be obtained from CPE phase angle using32–34)
Cp¼ðQ
p RpÞð1=npÞ
Rp ; ð7Þ
whereQp is the admittance andnp is defined as the phase angle of the CPE impedance independent of the frequency, which is also characterized by the degree of non-ideality in the behavior of a capacitor.30,35)TheCpvalues of the passive layers are listed in Table II. The table shows thatRpandRct increase, whereas Cp and Cdl decrease with increasing electric voltage (in the range of 300 to 800 mV). The lowCp and high Rp obtained at high potential indicates a thick passive layer formation assuming a fixed dielectric constant for the passive film. The Cu dissolution rate increased with
1E-6 1E-5 1E-4 1E-3 0.01 0.1
0.0 0.5 1.0 1.5 2.0 (d) 0.01 0.25 0.30 0.35 (3) (4) (2) (5) (6) (1) E (V vs Ag/AgCl) I (A/cm2) (1)DF0.14 SR50 (2)DF0.14 SR100 (3)DF0.14 SR200 (4)DF0.14 SR400 (5)DF0.14 SR600 (6)DF0.14 SR800
1E-5 1E-4 1E-3 0.01 0.1
0.0 0.5 1.0 1.5 2.0 0.01 0.25 0.30 0.35 (5) (6) (3) (2) (4) (7) (1) E (V vs Ag/AgCl) I (A/cm2) (c) (1) DF0.07 SR200 (2) DF0.14 SR200 (3) DF0.24 SR200 (4) DF0.35 SR200 (5) DF0.5 SR200 (6) DF0.7 SR200 (7) Electropolishing 0 200 400 600 800 0 500 1000 1500 2000 DF(psi) Cu ECMP rate (nm/min) EV(mV vs Ag/AgCl) / SR(rpm) Electrical Voltage Speed of Rotating 0.0 0.2 0.4 0.6 (b) Down Force 200 400 600 800 20 40 60 80 R2 = 0.9427 R2 = 0.9917 (a) Id (mA/cm 2 ) E (mV vs Ag/AgCl) ECMP Electropolishing
Fig. 2. (a) Linear relationship between current density (Id) and electric potential against a reference electrode (Ag/AgCl) during ECMP and
electropolishing. The conditions of the down-force and the rotation speed were shown in Table I. (b) Cu ECMP rate as a function of electric potential (EV), down-forces (DF) and speeds of the rotating (SR) in a H3PO4electrolyte, where the two variables were kept constant. (c) Potential dynamic curves of Cu disk electrode for various values of DF [(1) 0.07, (2) 0.14, (3) 0.24, (4) 0.35, (5) 0.5, (6) 0.7, and (7) 0 psi] in a H3PO4electrolyte. The disk was rotated at 200 rpm. (d) Potential dynamic curves of Cu disk electrode for various values of SR [(1) 50, (2) 100, (3) 200, (4) 400, (5) 600, and (6) 800 rpm] in a H3PO4electrolyte. DF was 0.14 psi. The potential was scanned at 20 mV/s (c) and (d).
decreasing charge-transfer time-delay, which is defined as Rct Cdl. In Table II, the charge-transfer time-delay greatly decreased when the electric potential was increased from 300 to 800 mV, indicating that the Cu dissolution rate increases with electric potential for Cu ECMP. At a low voltage (300 mV), the charge-transfer time-delay decreased slightly with increasing down force and rotation speed, indicating that the Cu dissolution rate also increases with mechanical force. The value of Rp decreases and Cp increases from 0.07 to 70 psi, implying that the passive
layer on the Cu surface becomes thinner. The same results were obtained for rotation speed. The thickness of the passive layer with a high down force or rotation speed can be continuously reduced at low potential, especially for high rotation speeds. The Cu removal rate may be limited to 800 nm/min at low electric potential, as shown in Fig. 2(b). However, the effect of electric potential is greater than that of mechanical force.
Since electric potential greatly affects the Cu removal rate, it was increased in the following experiment. Figure 4
3 6 9 12 16 0 1 2 3 4 (a) 13Hz 790Hz - Z im ( Ω cm 2 ) Z re (Ω cm 2 ) EV300 DF0.14 SR200 EV500 DF0.14 SR200 EV800 DF0.14 SR200 3 4 5 6 7 8 0.0 0.3 0.6 0.9 1.2 (b) 0.3Hz 27Hz - Z im ( Ω cm 2 ) Z re (Ω cm 2 ) EV300 DF0.14 SR200 EV300 DF0.24 SR200 EV300 DF0.7 SR200 3 4 5 6 7 8 0.0 0.3 0.6 0.9 1.2 (c) 0.3Hz 27Hz - Z im ( Ω cm 2 ) Zre (Ω cm2) EV300 DF0.14 SR100 EV300 DF0.14 SR400 EV300 DF0.14 SR800 5 10 15 0 2 4 Rp Rct Rs Cdl Cp Measurement - Z im ( Ω cm 2 ) Zre (Ω cm2) (d) Rs Rs+Rct+Rp Simulation Rs+Rct
Fig. 3. (a) Nyquist plots of Cu films polished at various electric potentials (EV) in H3PO4electrolyte, (b) Nyquist plots of Cu films polished at various down-forces (DF) in H3PO4electrolyte, (c) Nyquist plots of Cu films polished at various speeds of rotation (SR) in H3PO4electrolyte, and (d) simulation results from the proposed equivalent circuit diagram that was matched with the impedance data of EV800DF0.14SR200 in (a). Proposed equivalent circuit diagram shows the electrochemical characteristics during ECMP in the electrolyte, whereRsis the bulk solution resistance,Cdlis the double-layer
capacitance,Cpis the passive-layer capacitance,Rctis the charge-transfer resistance (associated with double layer), andRpis the passive resistance of Cu
film.
Table II. Element values of equivalent circuit in Fig. 3(d) required for the best fitting of impedance spectra in Figs. 3(a)–3(c).
Rs Rct Cdl RC Rp CPEp Cp Polishing conditions ( cm2) ( cm2) (F cm2) (s) ( cm2) Qp n p (F cm2) (1cm2) EV300 DF0.14 SR200 3.01 2.58 19.12 49.3 2.29 0.32 0.61 0.261 EV500 DF0.14 SR200 3.02 2.96 4.67 13.8 2.33 0.31 0.64 0.2521 EV800 DF0.14 SR200 3.46 3.85 2.7 10.4 6.85 0 0.79 0.0017 EV300 DF0.35 SR200 2.7 2.55 8.93 22.8 1.78 0.34 0.67 0.2622 EV300 DF0.7 SR200 2.55 2.56 7.24 18.5 1.65 0.46 0.68 0.4054 EV300 DF0.14 SR400 2.97 2.18 10.64 23.2 1.58 0.35 0.7 0.2672 EV300 DF0.14 SR800 2.87 1.4 11.02 15.4 0.77 0.36 0.87 0.2952 S.-W. Chen et al. Jpn. J. Appl. Phys. 51 (2012) 036504
shows the Nyquist plots of Cu films polished at a high electric voltage of 1800 mV for various down force values and rotating speeds compared to the plot obtained at 300 mV taken from Fig. 3. The fitted values of the circuit components in Fig. 3(d) with Fig. 4 are summarized in Table III. The charge-transfer time-delay decreased mark-edly when the electric potential was increased from 300 to 1800 mV, indicating that the Cu dissolution rate increased. Moreover, the values ofRctandRpincreased and that ofCp decreased with increasing electric potential, indicating that the thickness of the barrier film increases with increasing electric potential. At the high voltage of 1800 mV, with increases of down force and rotation speed, the charge-transfer time-delay decreased and the Cu dissolution rate increased. Rp decreased with increasing mechanical force, especially for rotation speed, implying that the thickness of the passive film decreased with mechanical force.
The ratio of the variation in Rp from the maximum (denoted as high-level) and minimum (low-level) in the range of each variable from Tables II and III was calculated using
Mp¼R
pðhigh-levelÞ
Rpðlow-levelÞ; ð8Þ
whereMev
p is the ratio ofRp at the highest potential toRp at the lowest potential at a fixed down force and rotation speed, and Mdf
p and Mpsr are the ratios of down force and rotation speed, respectively. Mev
p values of 7.21, 8.67, and 14.34 were obtained for DF0.14SR200, DF0.7SR200, and DF0.14SR800, respectively. Mdf
p values of 0.72 and 0.87 were obtained for EV300SR200 and EV1800SR200,
respec-tively. Msr
p values of 0.34 and 0.67 were obtained for EV300DF0.14 and EV1800DF0.14, respectively. The value of Mpev is much higher than those of Mpdf and Mpsr. This indicates that the increase of potential mainly affects the production rate of the passive film. The values of Mpev for DF0.7SR200 and DF0.14SR800 are higher than that for DF0.14SR200. The growth rate of the diffusion layer increased with increasing down force or rotation speed probably due to an increase in Id. The value of Mdfp for EV1800SR200 is higher than that for EV300SR200, indicating that the growth rate of the diffusion layer increased with increasing potential. The same result was found for Msr
p.
To investigate surface chemical species, XPS survey spectra of the Cu thin films after ECMP were obtained at various electric voltages, down force values, and rotation speeds, as shown in Fig. 5, for Cu 2p3=2spectra. Each of the Cu 2p3=2 spectra can be decomposed into three peaks at
0 5 10 15 20 25 0 2 4 6 8 10 893Hz 6Hz 4 6 8 0 1 2 - Z im ( Ω cm 2 ) Z re (Ω cm 2 ) EV 300 SR200 DF 0.14 EV1800 SR200 DF 0.14 EV1800 SR800 DF 0.14 EV1800 SR200 DF 0.7
Fig. 4. Nyquist plots of Cu films polished at various electric potentials (EV), down-forces (DF), and speeds of rotation (SR) in H3PO4electrolyte.
Table III. Element values of equivalent circuit in Fig. 3(d) required for the best fitting of impedance spectra in Fig. 4.
Rs Rct Cdl RC Rp CPEp Cp Polishing conditions ( cm2) ( cm2) (F cm2) (s) ( cm2) Qp n p (F cm2) (1cm2) EV300 DF0.14 SR200 3.01 2.58 19.12 49.3 2.29 0.3198 0.61 0.261 EV1800 DF0.14 SR200 3.66 3.05 2.04 6.2 16.52 0.0059 0.78 0.0031 EV1800 DF0.14 SR800 3.22 3.63 1.79 6.5 11.04 0.0023 0.95 0.0019 EV1800 DF0.7 SR200 3.59 3.66 1.6 5.9 14.31 0.0036 0.81 0.0018 940 938 936 934 932 930 928 926
Intensity (arb. units)
Binding Energy (eV) (d) EV1800 DF0.7 SR200 (c) EV1800 DF0.14 SR800 (b) EV1800 DF0.14 SR200 Cu(OH)2CuO Cu2O/Cu (a) EV300 DF0.14 SR200
Fig. 5. Cu 2p spectra of Cu films polished at various electric potentials (EV), down-forces (DF), and speeds of rotation (SR) in H3PO4electrolyte.
932:0 0:2, 933:2 0:3, and 934:1 0:5 eV, which are assigned to Cu/Cu2O, CuO, and Cu(OH)2, respectively. Figure 6 shows XPS spectra of the O 1s line for the Cu films after ECMP for various conditions. The broad peak of the O 1s spectrum can be decomposed into three peaks at 530:2 0:1, 532:2 0:3, and 531:3 0:1 eV, which are attributed to CuO, Cu(OH)2, and Cu2O, respectively. The major peak of the O 1s spectrum corresponds to Cu2O, which is in good agreement with the high-intensity of the Cu2O peak obtained from the Cu 2p3=2 spectrum in Fig. 5. Figure 7 shows the relationship betweenRpand the intensity ratio of Cu2O/Cu(II), ICu2O=CuðIIÞ, for various ECMP con-ditions, where Cu(II) includes Cu(OH)2and CuO.ICu2O=CuðIIÞ increases with increasingRp, indicating that theRp increase is due to the growth of the Cu2O oxide film. XPS spectra of P were also acquired; no phosphorus was found on the Cu film surface after ECMP and electropolishing using a H3PO4 electrolyte.36)
There appears to be a combination of two processes in Cu ECMP: (i) electrically enhanced metal dissolution and oxide diffusion layer growth and (ii) mechanical wear refreshing the Cu surface of the passive layer. For a constant down force and rotation speed, the Cu removal rate and the thick-ness of the Cu diffusion film increased with increasing electric potential, as shown in Figs. 2(b) and 3(a). The Cu diffusion layer formed partly due to Cu ions reacting with oxygen or hydroxyl. For a constant electric potential, the Cu removal rate increased slightly and Rp decreased with in-creasing mechanical force at low electric voltage (300 mV),
as shown in Figs. 2(b), 3(b), and 3(c). Rp decreased more significantly with increasing mechanical force at high electric voltage (1800 mV). The Cu oxide film is suppressed by mechanical force, but the effect of the mechanical force on the Cu dissolution rate is lower than that of the electric potential. Therefore, electric potential dominates the Cu dissolution rate at high potential whereas the mechanical force is dominant at low potential. This removal mechanism is greatly different from that observed for CMP or elec-tropolishing.
4. Conclusions
The effects of electric potential and mechanical force during ECMP on the dissolution of a Cu film in H3PO4 electrolyte were investigated. The increase of the Cu removal rate at low potential is due to the effect of mechanical force enhanced by electric potential. The increase of the Cu removal rate at high potential is due to the effect of electric potential enhanced by mechanical force. The potentiody-namic measurements show that high electric potential greatly increases Id, the dissolution rate, andRp. The effect of mechanical force is not as large as that of electric potential. The value ofMev
p is much higher than those ofMpdf and Msr
p. EIS and XPS results reveal that Rp is directly proportional toICu2O=CuðIIÞ, implying that Cu2O is the major oxide film during ECMP.
Acknowledgments
This work was supported by the National Science Council of Taiwan (Grant Nos. NSC 98-2622-E-006-049-CC2 and NSC 98-2221-E-006-082-MY3). Technical support from the Nation Cheng Kung University is also acknowledged.
1) R. Liu, C. S. Pai, and E. Martinez:Solid-State Electron.43 (1999) 1003. 2) T. F. Edgar, S. W. Butler, W. J. Campbell, C. Pfeiffer, C. Bode, S. B.
Hwang, K. S. Balakrishnan, and J. Hahn:Automatic36 (2000) 1567. 3) D. Leggett, D. Frye, R. Munzlinger, A. Millaway, J. Serda, P. Burke, M.
Evans, B. Kraus, and G. Leggett: PTI Seminars, Inc., I, 1.0 (1998). 4) P. C. Andricacos, C. Uzoh, J. O. Dukovic, J. Horkans, and H. Deligianni:
IBM J. Res. Dev.42 (1998) 567.
5) K. W. Chen, Y. L. Wang, L. Chang, S. C. Chang, F. Y. Li, and S. H. Lin: 535 534 533 532 531 530 529 528 527
Intensity (arb. units)
Binding Energy (eV) (d) EV1800 DF0.7 SR200 (c) EV1800 DF0.14 SR800 (b) EV1800 DF0.14 SR200 Cu(OH)2 Cu 2O CuO (a) EV300 DF0.14 SR200
Fig. 6. O 1s spectra of Cu films polished at various electric potentials (EV), down-forces (DF), and speeds of rotation (SR) in H3PO4electrolyte.
0.6 0.9 1.2 1.5 1.8 0 5 10 15 y = 13.416x - 5.7651 R2 = 0.9506 Rp (ohm-cm 2 ) ICu2O/Cu(II)
Fig. 7. Relationship betweenRpand the intensity ratio of Cu2O/Cu(II), ICu2O=CuðIIÞ, for various ECMP conditions.
S.-W. Chen et al. Jpn. J. Appl. Phys. 51 (2012) 036504
Electrochem. Solid-State Lett.7 (2004) G238.
6) D. Ernur, V. Terzieva, W. Wu, S. H. Brongersma, and K. Maex:
J. Electrochem. Soc.151 (2004) B636.
7) D. Ernur, S. Kondo, and K. Maex:Jpn. J. Appl. Phys.41 (2002) 7338. 8) N. Chandrasekaran, S. Ramarajan, W. Lee, G. M. Sabde, and S. Meikle:
J. Electrochem. Soc.151 (2004) G882.
9) P. R. Besser and Q. T. Jiang:MRS Proc.795 (2003) U1.1.
10) F. Q. Liu, T. Du, A. Duboust, S. Tsai, and W. Y. Hsu:J. Electrochem. Soc. 153 (2006) C377.
11) Y. J. Oh, G. S. Park, and C. H. Chung:J. Electrochem. Soc.153 (2006) G617.
12) J. Huo, R. Solanki, and J. McAndrew:Electrochem. Solid-State Lett.8 (2005) C33.
13) Y. Hong, D. Roy, and S. V. Babu:Electrochem. Solid-State Lett.8 (2005) G297.
14) A. Tripathi, C. Burkhard, I. I. Suni, Y. Li, F. Doniat, A. Barajas, and J. McAndrew:J. Electrochem. Soc.155 (2008) H918.
15) A. S. Brown:IEEE Spectrum42 (2005) 40.
16) T. M. Kung, C. P. Liu, S. C. Chang, and Y. L. Wang:Nanosci. Nanotechnol. Lett.1 (2009) 17.
17) S. C. Chang and Y. L. Wang:J. Vac. Sci. Technol. B22 (2004) 2754. 18) D. Padhi, J. Yahalom, S. Gandikota, and G. Dixit:J. Electrochem. Soc.150
(2003) G10.
19) D. Tromans and G. Li:Electrochem. Solid-State Lett.5 (2002) B5. 20) Y. Hong, D. Roy, and S. V. Babu:Electrochem. Solid-State Lett.8 (2005)
G297.
21) A. Tripathi, C. Burkhard, I. I. Suni, Y. Li, F. Doniat, A. Barajas, and J. McAndrew:J. Electrochem. Soc.155 (2008) H918.
22) P. C. Goonetilleke, S. V. Babu, and D. Roy:Electrochem. Solid-State Lett. 8 (2005) G190.
23) L. Economikos, X. Wang, A. Sakamoto, P. Ong, M. Naujok, R. Knarr, L. Chen, Y. Moon, S. Neo, J. Salfelder, A. Duboust, A. Manens, W. Lu, S. Shrauti, F. Liu, S. Tsai, and W. Swart:Proc. IEEE Int. Interconnect Technology Conf., 2004, p. 233.
24) D. Shirley:Phys. Rev. B5 (1972) 4709.
25) S. C. Chang, J. M. Shieh, C. C. Hung, B. T. Dai, and M. S. Feng:Jpn. J. Appl. Phys.41 (2002) 7332.
26) F. H. Giles and J. H. Bartlett:J. Electrochem. Soc.108 (1961) 266. 27) K. Kojima and C. W. Tobias:J. Electrochem. Soc.120 (1973) 1202. 28) E. S. M. Sherif, R. M. Erasmus, and J. D. Comins:J. Colloid Interface Sci.
311 (2007) 144.
29) S. Chaudhari, S. R. Sainkar, and P. P. Patil:J. Phys. D40 (2007) 520. 30) L. Larabi, O. Benali, S. M. Mekelleche, and Y. Harek:Appl. Surf. Sci.253
(2006) 1371.
31) K. Kojima and C. W. Tobias:J. Electrochem. Soc.120 (1973) 1026. 32) E. Barsoukov and J. R. Macdonald: Impedance Spectroscopy: Theory,
Experiment, and Applications (Wiley, Hoboken, NJ, 2005).
33) M. E. Orazem, P. Shukla, and M. A. Membrino:Electrochim. Acta47 (2002) 2027.
34) H. Ma, S. Chen, B. Yin, S. Zhao, and X. Liu:Corros. Sci.45 (2003) 867. 35) F. B. Growcock and R. J. Jasinski:J. Electrochem. Soc.136 (1989) 2310. 36) J. L. Fang and N. J. Wu:J. Electrochem. Soc.136 (1989) 3800.
