1 Introduction

Copper has been the main type of interconnect metal used for integrated circuits in devices with reduced dimensions [1, 2]. Chemical mechanical polishing (CMP) is a key enabler for global and local planarization and has been intensively investigated from various angles [3, 4]. According to the standard dual-damascene interconnect process for copper, the barrier layer (Ta/TaN, Mo, Co, Ru, and their alloy) is deposited onto the dielectric prior to the electroplating of copper [5, 6]. Therefore, it is necessary for the copper and barrier layers to be simultaneously planarized with the aim of gaining appropriate surface quality and material removal rate (MRR) during the last polishing step (barrier polishing) [7, 8].

Ruthenium (Ru) is becoming one of the most promising barrier layers as the feature size of chip elements continues to shrink below 14 nm [9–11]. Many oxidants such as iodate, hydrogen peroxide, ammonium persulfate, and periodate have been used in the Ru polishing process [12–15]. Among them, potassium periodate is the most widely used oxidant, and several investigations into the copper CMP process in periodate-based slurries have been carried out in recent years [16, 17]. These studies have mainly focused on the aspects like chemical and mechanical material removal, the surface quality, and the control of galvanic corrosion. Jiang et al. [18] studied the synergetic effect of benzotriazole and the nonionic surfactant Pluronic® P103 on copper in KIO4-based slurries. It was found that the addition of Pluronic® P103 in benzotriazole could produce a complete passivation film to protect copper from mechanical excessive abrasion. Chockalingam et al. [19] found that the presence of 2 mM BTA and 3 mM sucrose in 30 mM KIO4-based slurry could minimize the galvanic corrosion of Cu–Mn to 0.01 V at pH 10. Zeng et al. [7] obtained good material removal selectivity between copper and ruthenium to about one by controlling the concentration of glycine.

However, few studies have covered the chemical mechanical synergetic effects during copper CMP with periodate as oxidants. Copper CMP could be identified as a transient chemical process with intermittent mechanical phenomena. The total material loss can be described in the following equation [20]:

$$V = V_{\text{c}} + V_{\text{w}} + \varDelta V_{\text{c}} + \varDelta V_{\text{w}}$$
(1)

where V is the total material loss. V c is the pure corrosion without the influence of wear, and V w is the material loss due to wear without the influence of corrosion. ΔV c and ΔV w are the additional corrosion attributed to wear corrosion and corrosion wear effects, respectively. ΔV c and ΔV w could be considered as the synergetic effects on material removal during CMP. There are some studies available on the tribochemical of copper and other materials [21, 22]. Li et al. [23] investigated the mechanism of tribology phenomena interaction with film formation during the CMP process in a H2O2-based slurry. Stojadinović et al. [24] found that for Ru, the wear-accelerated corrosion effect is not that significant in the passive and transpassive domain. On this basis, researchers could further identify the most dominant factor in material loss during CMP under different working conditions. Several tribochemical models of copper have also been proposed, in which it was assumed that the transient passivation behavior and some characteristics of mechanical interactions are the most important constituents.

In this paper, the tribochemical behavior of copper in a KIO4-based slurry during CMP was investigated. No other additives existed in the slurries or solutions in order to eliminate other possible disruptive factors. Tribocorrosion tests were conducted to reveal the tribochemical behavior of copper. The changes in potential and current caused by sliding were studied in detail. The surface film composition of copper was analyzed by Raman spectroscopy. Finally, in the CMP chemical experiments, the proportion of the wear-accelerated corrosion to the total material loss caused by corrosion was calculated over a wide pH window of 4–11.

2 Experimental

2.1 Tribocorrosion Experiments

Copper (99.99 % purity), which acted as the lower sample as well as the working electrode during the experiments, was cut into 20 mm × 20 mm × 2 mm squares. Before the experiments, the copper squares were polished on a Struers bench top polisher (TegraPol-35) until the surface roughness was below 3 nm. Then, the samples were cleaned in DI water and dried in a N2 gas stream. The upper sample was Al2O3 ball with a diameter of 4.763 mm. The solution contained 0.015 M KIO4 as oxidant. The pH value was adjusted to 4, 6, 9, and 11 with KOH and H2SO4 as pH adjustors. The surface area of the copper sample exposed to the solution was 1.526 cm2.

The tribocorrosion experiments were conducted with a universal microtribometer (UMT-3, CETR) and a potentiostat (273A EG&G, Princeton Applied Research) in a three electrode electrochemical cell mounted on the tribometer. The reciprocating sliding tribometer has been shown in more detail elsewhere [23]. A platinum electrode was used as the counter electrode, and a Ag/AgCl (3.5 M) electrode was used as the reference electrode. During the experiments, the Al2O3 ball did reciprocating sliding against the copper surface. The sliding distance was 2 mm, the frequency was 4 Hz, and the applied force was 6 N. The sliding duration was 100 s, and the interval between slidings was 300 s. Firstly, variations in the open-circuit potential (OCP) were measured with and without sliding. Then, changes in current were continuously monitored under the OCP.

2.2 CMP Chemical Experiments

In order to study the steady-state electrochemical behavior of copper during CMP, CMP chemical experiments were carried out. The apparatus was modified on an ultra-low pressure CMP machine. The schematic diagram of the CMP chemical apparatus is shown in Fig. 1. The copper sample (99.99 %) was 12.5 mm in diameter and about 1 mm in thickness. An IC1010/Suba IV pad was used during the polishing process. The slurry used contained 0.015 M KIO4 without any abrasives. The pH was adjusted to 4, 6, 7, 9, 10, or 11, respectively. Potentiodynamic polarization experiments were conducted on an electrochemical workstation (CHI600C, Chenhua) at a scan rate of 1 mV/s. The counter electrode was platinum, and the reference was a Ag/AgCl (3.5 M) electrode. During the polishing process, the applied force on the copper sample was 2.9 psi. The copper sample was stationary, and the pad rotated at a speed of 0 or 100 rpm.

Fig. 1
figure 1

Schematic diagram of the instrument used for the CMP chemical experiments

Full size image

2.3 Surface Analysis

The surface morphologies of the copper after tribocorrosion tests were observed using a Veeco MicroXAM optical interferometer. The samples were cleaned in DI water and dried in a N2 gas stream after the tribocorrosion tests. To characterize the surface film composition, Raman spectra were obtained to analyze the surface reaction products inside and outside the wear track. The Raman spectra were measured using a Horiba Jobin–Yvon Raman spectrometer with a HeNe laser (632.8 nm).

3 Results

3.1 Tribocorrosion Experiments

Figure 2 shows the surface morphology of the unworn surface after the experiments. It can be seen that in acidic solutions, the copper surface undergoes severe corrosion. The surface roughness (Ra) is listed in Table 1. The results correspond well with the surface morphology in Fig. 1. Under acidic conditions, surface roughness is obviously high, but it remains comparatively low (about 13 nm) under alkaline conditions.

Fig. 2
figure 2

a–d Optical microscope picture of the copper unworn surface (×200) after tribocorrosion experiments; e–h surface micrographs by white light inteferometer; a and e are at pH 4; b and f are at pH 6; c and g are at pH 9; d and h are at pH 11

Full size image
Table 1 Surface roughness of the unworn copper as a function of solution pH
Full size table

Figure 3 shows the typical wear track morphology after reciprocating sliding. It can be seen that under acidic conditions, the wear track exhibits a rough surface while it has smooth sidewalls under alkaline conditions (Fig. 3b). The wear track depth of the copper was extracted from the cross-sectional profile. The wear track depth is shown in Table 2. It is obvious that the scratch depth is greatly affected by the pH value, and it increases as the solution becomes more acidic.

Fig. 3
figure 3

a Typical scratch morphology at pH 6, b Cross-sectional view of the scratch and its depth in tribocorrosion experiments at pH 4 and 11

Full size image
Table 2 Scratch depth of the wear track as a function of solution pH
Full size table

The change in the OCP is shown in Fig. 4 for different values of the solution pH. The arrows in the picture indicate the commencement of the rubbing in reciprocationg sliding process. The OCP decreases with an increase in the pH value. Under alkaline conditions, the obvious change in the OCP to lower values is caused by rubbing, while this trend is unobservable under acid conditions. At the beginning of the rubbing, the OCP sharply jumps to a lower value compared with the initial OCP (pH 9 and 11). However, the OCP evolution during the rubbing process is not the same. At pH 9, the OCP progressively decreases, while for pH 11 it remains almost unchanged at a specific value during the whole rubbing process.

Fig. 4
figure 4

Changes in the OCP with time before, during, and after rubbing in KIO4-based solutions for tribocorrosion experiments at pH 4, 6, 9, and 11

Full size image

An increase in current during rubbing is observed in Fig. 5. The rise in cathodic current is evident in both the acidic and alkaline solutions. The current increase becomes increasingly larger when the pH value is decreased. At the end of the rubbing periode, the mechanical abrasion ceases and the current recovers the initial value. It is observed that after each rubbing process, the drop in the current for the acidic solutions is slower and smoother than that for the alkaline solutions.

Fig. 5
figure 5

Changes in the current with time before, during, and after rubbing in KIO4-based solutions for tribocorrosion experiments at pH 4, 6, 9, and 11

Full size image

3.2 CMP Chemical Experiments

The corrosion current density with and without polishing is shown in Fig. 6. Over the entire pH range (pH4-11), the current density without polishing (I c) is lower than that (I cc) during polishing. However, the two curves in Fig. 6 follow the same tendency: The current density is at its minimum when pH is near neutral, while it increases when the solution is either more acidic or more alkaline. At pH 4, the current reaches the maximum (~21 and ~29 μA for the static and polishing condition, respectively), which is significantly larger than that of other pH conditions. If the difference between I cc and I c at a certain pH value is denoted by ΔI c, the wear-accelerated corrosion effect could be evaluated by the ΔI c/I cc ratio. The corrosion potential during polishing and in static condition are denoted by E cc and E c, respectively, and ΔE corr indicates the corrosion potential difference. The relevant parameters are listed in Table 3. It is clearly seen that the corrosion potential during polishing rises compared with that in static condition. With respect to the wear corrosion effect, ΔI c/I cc reaches its maximum when the solutions are weakly alkaline (0.5428 for pH 9 and 0.5889 for pH 10). This trend is plotted in Fig. 7.

Fig. 6
figure 6

Corrosion current density of copper before (0 rpm) and during polishing (100 rpm) as a function of pH in CMP chemical experiments

Full size image
Table 3 Parameters in the CMP chemical tests
Full size table
Fig. 7
figure 7

ΔI c/I cc and ΔE corr during CMP as a function of pH in CMP chemical experiments

Full size image

Figure 8 shows the typical dynamic polarization curve during polishing and under the static conditions in the CMP chemical experiments (pH 9). The Tafel curve in the polishing state undergoes slight oscillation compared with that in the static state. The cathodic Tafel slope (222 mv/dec) under the polishing condition is apparently larger than that under the static condition (179 mv/dec). The manner in which the polishing process affects the corrosion of copper could be inferred from the data mentioned above.

Fig. 8
figure 8

Potentiodynamic polarization curves in static and polishing condition at pH 9 in CMP chemical experiments

Full size image

3.3 Raman Spectra Experiments

The copper polishing in KIO4-based slurry is quite different from that in the H2O2-based slurry. Apart from copper oxide and copper peroxide reported in early literature, the surface passivation film also contains copper iodate and copper periodate, both of which are insoluble in water [25–27]. Li et al. [28] found that during the CMP process in iodate-based slurry, the copper surface was passivated with copper oxides/Cu(IO3)2. Similar experimental results were observed by Qiuliang Luo and M. ANIK [29, 30]. In the periodate-based slurry, the composition of the passivation film on copper surface is more complex [31]. Jiang et al. [18] believed that the copper surface passivation film contains not only Cu(IO3)2 but also some Cu(IO4)2. However, the reaction products on copper with periodate as oxidant are quite complex. The surface reaction products not only contain copper oxides and copper iodate, but also copper iodides, iodine, and copper periodates [32, 33]. However, the copper periodates exist in different forms and could not simply be described as Cu(IO4)2 [34]. The periodate reaction compound existing on copper may contain the copper orthoperiodate hydrate (Cu2HIO6·nH2O), CuH4I2O10·6H2O etc [31, 35, 36]. All the insoluble compounds along with copper oxides are integrated into the passivation film. The redox products formed on the copper surface are due to the oxidation of copper and the reduction of periodate, because the IO4 −/IO3 − equilibrium lines are located at higher potentials than those of Cu2+/Cu and Cu2O/Cu according to the Pourbaix diagram [30]. The main reactions could be described in the following equations:

$$\rm{IO_{4}^{ - } + H_{2} O + 2e^{ - } \to IO_{3}^{ - } + 2OH^{ - }}$$
(2)
$$\rm{ Cu + H_{2} O \to Cu_{x} O + ne^{ - } \, (x = 1 \, or \, 2, \, n = 2 \, or \, 1) }$$
(3)

The composition of the copper surface passivation film was analyzed by Raman spectroscopy. The unworn surface and the wear track were each measured. More than one kind of copper oxide may exist on the surface, i.e., CuO, Cu2O, and Cu(OH)2 [26, 37]. There are often several frequency peaks for a certain material because of the various vibration forms and their overlap. For CuO, the peaks in the Raman spectra are mainly located at the wave numbers 160, 250, 330, 410, and 635 cm−1 [27, 38]. The Raman peak associated with the Cu-OH stretch is assigned to Cu(OH)2, which often covers a wide range of wave numbers between 450–470 and 540–580 cm−1 [39]. The most commonly reported peak for the Cu-OH vibration is 488 cm−1 [38, 40]. The peaks located at 220, 588, 633, and 645 cm−1 indicate the existence of Cu2O [41–43]. The peaks at 124 and 147 cm−1 can be assigned to Cu+(I−)OH− complex, and the peak at 124 cm−1 is assigned to Cu-I vibration in CuI [44]. The Raman spectra for iodine are located at 180 and 195 cm−1 [32, 45]. The Raman spectra results may vary little (10 cm−1 error) due to the preparation of the sample and the experimental conditions.

For Cu(IO3)2, the reported Raman spectra data are scarce. It has been reported that for the anhydrated copper iodate, the peaks are located at 697, 737, and 797 cm−1 in the 550–4,000 cm−1 region. However, for hydrated Cu iodate [3Cu(IO3)·2H2O], the frequency locates at 720, 745, and 767 cm−1, which are obviously different to that of the anhydrated form. In the lower wave number region of 35–550 cm−1, there exist many peaks with strong vibrations in the absorbance curve, which has been reported by Nassau et al. [46]. With regard to copper periodate, as mentioned previously, complex reaction products are formed in the periodate-based slurry; thus, the Raman spectra are hard to analyze. M. Botova et al. [47] found that for CuH4I2O10·6H2O, the frequency peaks were located at 620–630, 671, and 704 cm−1.

Figure 9 shows the Raman spectra of the wear track in the tribocorrosion experiments. It can be seen that when the pH value decreases, the Raman spectra intensity increases, which indicates higher content of the reaction products. However, the peak position for all the pH values is almost the same. The main reaction products on the copper surface are Cu2O, CuO, Cu(OH)2, and CuI. Copper periodate, copper iodate, and iodine could be barely seen on the rubbed surface. Figure 10 displays the Raman spectra of the unworn copper surface in the tribocorrosion experiments. It is obvious that the condition varies distinctively as a function of pH. The Raman spectra density under acidic conditions is higher than that under alkaline conditions, which corresponds well with that of the wear track mentioned above in Fig. 9. It can be seen that at pH 6 (Fig. 10b), there is an extremely high gauss peak indicating the obvious existence of CuH4I2O10·6H2O at about 678 cm−1. The iodine peaks emerge when the solution is acidic only, while for the entire pH values, CuI could form on the corroded copper surface in periodate solutions. The peak for Cu2O in acid conditions is comparatively stronger than in acid solutions, which is in keeping with the relevant research results [25]. To sum up, the surface passivation film mainly consists of copper oxides, copper hydroxide, and copper iodide. Small amounts of copper iodate, copper periodate, and iodine can exist under specific pH conditions, especially there forms evident copper periodate on the copper surface when the solution pH is 6 and I 2 appears only when the solution is acidic, which is clearly indicated in Fig. 10.

Fig. 9
figure 9

Raman spectra of the wear track at pH 4, 6, 9, and 11

Full size image
Fig. 10
figure 10

Raman spectra of the unworn surface for a pH = 4, b pH = 6, c pH = 9, and d pH = 11. The numbers in red, black, yellow, blue, purple, green, gray, and pink color represent Cu2O, CuO, Cu(OH)2, Cu(IO3)2, 3Cu(IO3)2·2H2O, CuH4I2O10·6H2O, CuI, and I 2 peaks, respectively (Color figure online)

Full size image

4 Discussion

4.1 Corrosion Behavior of Copper in Periodate-Based Solution

For static etching of copper, it could be conducted that the copper surface undergoes severe corrosion in acidic solutions and the corrosion is much mitigatory in alkaline solutions. Obvious pitting corrosion could be seen on the copper surface at pH 6 and pH 4 (Fig. 1). The corrosion current densities measured in the CMP chemical and tribocorrosion experiments may also support this opinion. The static current density under OCP, together with the current density calculated from Tafel plots, could indicate the corrosion behavior to some extent [21, 24, 48]. Figure 6 shows that without polishing, the corrosion current density in acid conditions is much higher than that under alkaline conditions, which corresponds well with the current density measured between sliding intervals at different pH values (Fig. 5). The strong corrosion under acidic conditions could be attributed to the three main reasons below. Firstly, the oxidizability of IO4 − in acidic solutions is higher than that in alkaline solutions [30, 49], as a result of which corrosion is much more severe at pH 4 and 6 than at higher pH. Secondly, the incomplete passivation in acidic solutions could cause local galvanic corrosion. As is reported in literature, the surface oxide passivation film is incomplete and the copper only has patches (islands) of Cu2O and no CuO at low oxidizer concentration [25, 50]. In addition, the Raman spectra in Fig. 10 indicate the obvious existence of I 2, Cu2O, and CuH4I2O10·6H2O on copper surfaces at pH 4 and pH 6. The formation of copper iodate and copper periodate is more likely to occur on the copper surface in acidic solutions [18, 34]. Therefore, the patches of passivation film will contribute to the local galvanic corrosion, thus pitting will happen. Thirdly, it is reported that the copper iodate and copper periodate have a high solubility product and this could enhance the CMP MRR [28, 35, 51]. Therefore, the comparatively high solubility of the surface products could also enhance the static chemical corrosion in acidic solutions.

4.2 Tribocorrosion Behavior of Copper in Periodate-Based Solution

When abraded, the copper surface OCP is greatly affected (jumps to a lower value) at pH 9 and pH 11. The rubbing process in the experiments could expose more fresh copper surface to the solutions, improving the redox reaction. However, as discussed above, the copper surface in acidic solutions is severely corroded and the area of the fresh surface produced in the scratching process is negligible. Copper in the alkaline solutions is more sensitive to rubbing process because the surface is well passivated, mainly with CuO/Cu(OH)2/Cu2O/CuI (Fig. 10d, e). The change in OCP is due to the coupling of the fresh surface with the passivation film. At the beginning of the rubbing, the sharp decrease in OCP is because of the breakdown of the passivation film. During the rubbing process, the OCP progressively becomes more negative at pH 9, but it remains constant at pH 11. This could be explained by the galvanic coupling models arising during tribocorrosion [21, 52]. As a consequence, it could be inferred that at pH 9, the galvanic coupling between the wear track and unworn surface dominates, whereas the galvanic coupling within the wear track dominates at pH 11.

For the currents under the OCPs during rubbing process (Fig. 5), the initial sharp drop in current at pH 9 and pH 11 should be due to the large cathodic current of the unworn surface because the potential becomes more negative, which is clearly shown in Fig. 4 [53]. The similar phenomenon could not be seen at pH 4 and 6 because the OCP remains almost constant both with and without rubbing. The increase in current during rubbing should be due to the continuous expansion of the wear track surface and the galvanic corrosion, which has been discussed above [23]. The large increase in the currents at pH 4 and 6 is mainly attributed to the increase in the wear volume, which is shown in Table 2. However, for pH 6 and pH 4, the increase in current is comparatively ruleless. The coarse sidewalls of the wear track in the acidic solutions (Fig. 3b) suggest that the severe corrosion results in inhomogeneous wear track surfaces, which could lead to the random growth forms of the currents. The regular increase in the current in alkaline solutions at pH 9 and pH 11 should primarily be due to the galvanic corrosion and expansion of the wear track, which has been discussed in previous paragraph.

When the rubbing process ends, the passivation behavior of the copper depends on the different pH values. In acidic solutions, the absolute value of the current decreases more slowly than in alkaline solutions. This fact suggests that in acidic solutions, the passivation of copper is distinctly slower than in alkaline solutions.

The wear track surface film composition does not vary much at the different pH values. It could be inferred that in alkaline solutions, the copper surface is prone to be covered with a thin, complete cupric, and cuprous passivation film. However, in acidic solutions, the film is comparatively thick, but not compact enough. The surface film is mostly composed of copper oxides, copper iodide, and copper hydroxide. Copper iodate, copper periodate, and iodine are negligible compared with that on the unworn surface, and this could be explained by the fact that the insoluble compounds are easy to be removed, which has been discussed in Sect. 4.1.

4.3 Wear-Accelerated Corrosion Mechanism During Copper CMP in Periodate-Based Slurry

Most researchers investigate the electrochemical mechanism of metals during abrasion by conducting tribocorrosion tests [54, 55]. However, the rubbing process could not completely simulate the polishing process. For instance, the applied load is high and the scratch depth of the surface is large (Fig. 3b) compared with the copper surface film removal rate during CMP (~100 nm/min). Therefore, the CMP chemical experiments could be a better method for representing the electrochemical mechanism during CMP conditions. Figure 8 shows that the polishing process greatly affects the cathodic branch of the Tafel plot, which results in higher corrosion potential and corrosion current density. The asperities of pad could remove the surface passivation film of copper, and bare copper surface would be exposed to the solution, which could afford more cathodic reaction sites for the redox reaction (assuming no galvanic corrosion considered) [14]. The cathodic Tafel slope is quite large (−222 mV/dec), which means that the reaction is not completely electrochemically controlled but diffusion controlled to some extent.

Wear-accelerated corrosion and pure corrosion are responsible for the overall material loss due to corrosion [37]. The MRR due to wear corrosion (r Δc) and corrosion (r c) could be calculated based on Faraday’s law:

$$r = iM/n\rho F$$
(4)

where i is the corrosion current, F is the Faraday constant, n is the valence of copper in the research system under, ρ is the density, and M is the atomic mass of copper. It could be inferred from Eq. 4 that the ΔI c (Table 3) could reflect the wear corrosion effect in the CMP process of copper in a periodate-based slurry. ΔI c/I cc refers to the proportion contributed by the wear corrosion effect to the total corrosion induced material loss. It is clearly shown in Fig. 7 that the wear-accelerated corrosion is significant in weakly alkaline solutions (at pH 9 and 10) and it contributes the least at pH 6. Based on the analysis mentioned above, it is supposed that the wear-accelerated corrosion is due to the following two factors: exposure of the fresh copper to slurry during the abrading of the copper surface film and the local galvanic corrosion caused by coupling between the depassivation and the passivation microregions.

5 Conclusions

The tribocorrosion behavior of copper in periodate-based slurry has been studied in this paper. The tribocorrosion characteristics have been investigated by tribocorrosion experiments. Raman spectra experiments have been conducted to analyze the surface film composition. In order to explore the wear-accelerated corrosion effect in copper CMP with periodate as the oxidant, CMP chemical experiments have been done to calculate the ratio of ΔI c/I cc. The experimental results are beneficial for further optimizing the slurry composition and the operation conditions of copper CMP. The following conclusions have been drawn:

  1. 1.

    In periodate-based solution, copper static corrosion varies significantly as a function of solution pH. The corrosion becomes more severe at lower pH values. The reasons for the serious corrosion in acidic solutions could be due to the stronger oxidability of periodate, the comparatively high dissolution of surface reaction products, and the local galvanic corrosion.

  2. 2.

    The chemical and electrochemical reaction products of the copper surface in periodate-based solutions are complex. The surface passivation film mainly consists of copper oxides, copper hydroxide, and copper iodide. Small amounts of copper iodate, copper periodate, and iodine can exist under specific pH conditions, especially there forms evident copper periodate on the copper surface when the solution pH is 6. I 2 appears only when the solution is acidic. Abrasion could help to get a uniform passivation film on copper surface consisting of copper oxide, copper hydroxide, and copper iodide only. The surface passivation film on copper in alkaline conditions is denser, more complete than that under acidic conditions, which could better inhibit the corrosion of copper.

  3. 3.

    Copper is more sensitive to scratching under alkaline conditions than under acidic conditions. The wear-accelerated corrosion is mainly due to the exposure of more cathodic reaction sites to the slurry for participation in the redox reaction and to the local galvanic corrosion during CMP process. The repassivation process of copper in acidic solutions is much slower than that in alkaline solutions.

  4. 4.

    During the CMP process, wear-accelerated corrosion plays a significant role in total corrosion material loss in weak alkaline slurries (at pH 9 and pH 10). However, the wear corrosion effect is the weakest at a solution pH of 6. The high wear corrosion proportion of the total corrosion (ΔI c/I cc) could help to obtain a better surface quality and acceptable MRR during CMP.