1. Introduction

With the ongoing development of interconnects toward the nano-scale, increase in the degree of device integration, and the evolution of multiplayer interconnects on chips, copper has been used to replace aluminum as the interconnect material in integrated circuits (IC) because of its excellent intrinsic electromigration resistance and a lower resistivity. This contributes to a greatly reduced resistance--capacitance (RC) delay^[1-5]. The integration of copper into an IC manufacturing process can be implemented by using the dual damascene technique, in which chemical mechanical planarization (CMP) has become an essential step that makes the integration of copper as interconnect metal feasible^[6-14]. However, the CMP process also leaves lots of residual contaminants such as organic residues, abrasive particles and metallic contaminants on the copper surface that can degrade the electrical properties of ICs, lower the conductivity of Cu and lead to poor adhesion of the subsequent layers^{[15, 16]}. Organic residues may be the primary source of defects and mainly originate from benzotriazole (BTA), often used as an inhibitor to achieve good surface planarization because of the formation of non-soluble BTA complexes^[17-20].

Recently, several reports demonstrated that alkaline-based post Cu CMP cleaning solutions are preferred over acidic solutions since they can remove organic residues better^[21-25]. Venkatesh et al.^[23] proposed an alkaline solution consisting of tetra methyl ammonium hydroxide (TMAH) as the cleaning agent and arginine as the chelating agent. It was reported that the proposed cleaning solution showed good ability in removing BTA from the copper surface and also yielded a lower surface roughness. Manivannan et al.^[22] developed a non-amine-based alkaline cleaning solution, cesium hydroxide (CsOH) and potassium hydroxide (KOH) were used as cleaning agents and ethylene glycol was used as a corrosion inhibitor; they suggested that both solutions exhibited high BTA removal efficiency. Our previous work has developed a novel alkaline chelating agent and a nonionic surfactant for BTA and colloidal silica removal^{[26, 27]}. However, the exact role of each compound was not reported and the synergetic effect of chelating agent and surfactant has not been comprehensively studied. In this work, the wetting ability of the proposed chelating agent was first studied. Then the synergetic effect of chelating agent and nonionic surfactant on BTA removal was investigated, and the corresponding polishing mechanism was also revealed.

2. Experiment

2.1 Wetting ability

The wetting ability of cleaning solutions on the copper surface was investigated by the contact angle of the experiment. The contact angle of Cu surface was measured using a JC2000D contact angle analyzer (purchased from Shanghai Zhongchen Limited). Copper wafer coupons of 4 $\times$ 4 cm² used for contact angle measurements were cut from 300 mm blanket copper wafers (purchased from SKW Associates, Inc.). The Cu wafers were pre-cleaned in iso propyl alcohol (IPA) and dried by blowing N₂ gas before the wetting ability test. The FA/O chelating agents used in this experiment was purchased from Tianjin Jingling Microelectronic Material Limited, pH=13.85, described in detail in Reference ^[26]. Primary Alcohol Ethoxylate ((C₁₂H₂₅O $\cdot$ C₂H₄O)_n) has been used as the nonionic surfactant in all these experiments. The composition of the cleaning chemistry used for all experiments is shown in Table 1.

2.2 Electrochemical measurements

BTA contaminant removal from copper coupons was characterized using a CHI600C electrochemical workstation (Shanghai Zhongchen Digital technique apparatus Co., Ltd.) in a standard three-electrode. The electrochemical measurements were conducted in 250 mL electrochemical cells equipped with a saturated calomel reference electrode (SCE) and a Pt counter electrode. Cu coupons (1 $\times$ 1 cm², 99.99% pure) were first treated with a 1 wt% BTA solution to form Cu-BTA working electrode prior to each experiment. The potentiodynamic polarization curves were obtained in the range of $\pm$ 500 mV with respect to the open circuit potential (OCP), at a scan rate of 5 mV/s. Each experiment was repeated three times and the average and standard deviations are reported.

2.3 CMP and post-CMP cleaning

300 mm Reflexion-LK module CMP tool from Applied Materials Inc (AMAT). was used in this experiment. The polish step was performed with the same consumables and polishing conditions across all experiments. The post-CMP cleaning is composed of megasonic tank, 1st brush module, 2nd brush module and isopropyl alcohol (IPA) dryer. The schematic of the cleaning module is show in Figure 1. Cleaning solutions incorporated with a polyvinyl alcohol (PVA) brush were applied in the two brush modules sequentially. 300 mm copper patterned wafers based on 55 nm feature size were used to evaluate the cleaning performance of the proposed cleaning solution. The defects scan and review monitoring were performed by KLA Tencor 2825 bright field inspection system and AMAT SEMVision G4 review system, respectively.

3. Results and discussion

3.1 Wetting ability

Wettability of Cu surface is an important aspect in post CMP cleaning. It has a direct impact on the cleaning results as a hydrophilic surface enables relatively easy flushing away of contaminants and minimizes watermarks. Wettability studies usually involve the measurement of contact angles as the primary data, which indicates the degree of wetting when a solid and liquid interact. Small contact angles correspond to high wettability, while large contact angles correspond to low wettability. The images of a sample contact angle measurements are shown in Figure 2, which shows the droplet of deionized water (DIW) and solution C on copper surface; the highly hydrophilic copper surface could be helpful to achieve an efficient post CMP cleaning for organic contamination removal. As shown in Figure 3, the contact angle of water on a fresh copper surface is nearly 30^°, which shows it is hydrophilic, and it was taken as reference for this study. The contact angle was decreased to $\sim$ 22^° by adding 150 ppm FA/O chelating agent, while with further addition of 100 ppm of nonionic surfactant into solution A, the contact angle reduced about to 12^°. When the concentration of nonionic surfactant improved to 200 ppm, the contact angle decreased even further to 8^°; the results indicate that nonionic surfactant is more effective in making the surface hydrophilic. However, in view of the comparison results of C and E or B and D, FA/O chelating agent has a negligible effect on altering hydrophilicity of the solution when the nonionic surfactant is present.

The Cu surface was dipped in 1 wt% of BTA solutions to create adsorbed BTA film on Cu and simulate the effect of BTA adsorption from barrier slurries during CMP, then BTA-treated Cu samples were subsequently cleaned by various types of solutions for 1 min and the contact angle was measured and the results are presented in Figure 4. The experiment showed that the contact angle of a water drop on the Cu-BTA surface is about 64^°, the contact angle of Cu-BTA coupons cleaned by only DIW was still remaining high, which shows that the PVA brush incorporated with DI water is not an efficiency way for BTA removal. The addition of 150 ppm FA/O chelating agent (Solution A) reduces the contact angle to about 35^° and further addition of 100 ppm of nonionic surfactant (Solution B) decreases the contact angle even further. As the nonionic surfactant increased to 200 ppm, the contact angle significantly decreased to 28^°. All those results reveal that FA/O chelating agent is more efficient in removing Cu-BTA polymer, the addition of nonionic surfactant improves the capability of removing BTA. Compared with solutions D and B, both of the solutions have the same concentration of nonionic surfactant (100 ppm). While the concentration of chelating agent of D and B is 75 and 150 ppm respectively, it is seen from Figure 4 that the CA of Cu-BTA coupons cleaned by solution D is higher than solution B; this reveals that more chelating agents can improve the capacity of cleaning solution for BTA removal. By comparison, the lowest CA of Cu-BTA coupons was obtained by using solution E, and this reveals that more chelating agent combined with more nonionic surfactant can achieve good capability for BTA removal in the experimental composition range.

Potentiodynamic polarization curves of the copper treated with 1wt% BTA after treating with the various cleaning solutions are studied, and the results are shown in Figure 5. Anodic and cathodic branches of the polarization plots were extrapolated, and the corrosion current density ( ${I_{{\rm{corr}}}}$ ) was obtained from the intersection of the tangent lines. The ${I_{{\rm{corr}}}}$ values are tabulated in Table 2. Potentiodynamic polarization studies were conducted in the potential range of $\pm$ 0.3 V versus OCP, but only the linear region of the Tafel curve was considered for ${I_{{\rm{corr}}}}$ estimation. The corrosion current of BTA treated Cu cleaned in DI water alone is also included as reference. In the case of the reference, the BTA is not removed and hence corrosion current density is very low, the ${I_{{\rm{corr}}}}$ value of the BTA-treated copper coupon after treating with DIW was found to be 0.561 $\mu$ A/cm², which shows BTA passivation film remained on the copper surface. The corrosion current density of the copper coupon treated with 1 wt% BTA after treating only with 150 ppm FA/O chelating agent (Solution A) showed a higher ${I_{{\rm{corr}}}}$ of 3.967 $\mu$ A/cm². These results confirm that the FA/O chelating agent is a major contribution to the removal of BTA. It also can be found that the ${I_{{\rm{corr}}}}$ value increases from 3.967 to 5.055 $\mu$ A/cm² when the concentration of nonionic surfactant increases from 0 to 100 ppm, and continues to increase to 6.383 $\mu$ A/cm² when the concentration further increases to 200 ppm, which reveals that the addition of nonionic surfactant can promote the ability of the cleaning agent for BTA removal. From the results of comparison between solution A and solution D, it can be seen that the ${I_{{\rm{corr}}}}$ value of solution D is comparable to solution A, although the concentration of chelating agent is higher in solution A. It indicates that the combination of 75 ppm FA/O chelating agent and 100 ppm nonionic surfactant can effectively remove BTA, the synergetic effect of chelating agent and nonionic surfactant play an important role in the capability of cleaning chemistry for BTA removal.

300 mm copper patterned wafers based on 55 nm feature size were used to evaluate the cleaning performance of the proposed cleaning solution. The CMP process and cleaning process are fixed in the same condition. The defects scan and review monitoring were performed by KLA Tencor 2825 bright field inspection system and AMAT SEMVision G4 review system, respectively. Figure 6 shows total defect count (TDC) on the 300 mm copper patterned wafers treated with various cleaning solutions after Cu CMP. The TDC were including silica particles contamination, BTA residual, organic contamination, corrosion, copper oxide, damage and so forth. It is seen from Figure 6 that the defect count was very high after rinsing with only deionized water (DIW), the defect count reaches 21281 ea. However, the TDC significantly reduced to below 1000 when cleaning solution A was applied in the post Cu CMP cleaning process, which contains only 150 ppm chelating agent. The results reveal that small amounts of FA/O chelating agent can effectively remove the particles contaminated on the copper patterned wafer surface. As the content of nonionic surfactant in the cleaning solution increases, TDC is gradually decreased and it has minimum TDC for solution C. The solution D has nearly the same TDC compared to solution A; all the result are basically in accordance with the wetting ability test and electrochemical measurements results except for solution E.

To test the effectiveness of different cleaning chemistries in removing BTA residual, defects were identified with SEM and a typical image of defects was shown in Figure 7. In this case, BTA residual is separated out as an independent defect category. ''Other defects" refers to residual abrasive particles, debris from Cu oxide, organic contamination, damage, corrosion etc, and the relative defect density of BTA residual and other defects were shown Figure 8. Figure 8(a) shows that the wafer cleaned with only DIW has a large number of BTA counts (up to 13832 ea); by comparison, Solution A just has 271 ea of BTA residual, this reveals that the cleaning solution containing only 150 ppm chelating agent is working more efficiently for BTA removal. As seen from Figure 8(b), as the concentration of nonionic surfactant increased from 0 to 200 ppm, the defect counts of BTA residual was reduced from 271 to 81ea; this result demonstrated that the addition of nonionic surfactant into cleaning solution further improved the ability for BTA residual removal. Compared with solution A, solution D has better capability for BTA removal. The BTA residual on wafer surface after cleaning by using solution A is 271 ea, although the concentration of chelating agent in solution D is lower than A, the BTA counts for solution D is 107 ea, this result indicates that 75 ppm chelating agent combined with 100 ppm nonionic surfactant has better capability for BTA removal than only 150 ppm chelating agent. From Figure 8(b), we also find that copp er wafer cleaned by using solution E has the fewest amounts of BTA residual, it is reduced to about 57 ea. To our surprise, the total defect amounts of solution E is higher, it may be composed mainly of organic contamination of chelating agent, the example of this kind of defect was shown in Figure 7(a), and the count of each type (i.e., SiO₂ abrasive, CuO, particle, etc) at different cleaning solutions will be studied in the future research.

In acidic environments benzotriazole is present mainly in the undissociated form as BTAH ( $BTAH_2^ +$ =BTAH + H⁺), while in alkaline environments the BTA is present mainly in the BTA^- (BTAH=BTA^-+ H⁺). Due to the strong interaction between the nitrogen atom of BTA^- with the surface Cu atoms, the BTA molecule adsorbs at the Cu surface and forms a surface polymer complex of [Cu-(BTA)]_n. The chelating agent used in this study has an effect on Cu-BTA contamination removal. FA/O II chelating agent can complex Cu+/Cu²⁺ from Cu(I)BTA/Cu(II)BTA and cleave the Cu-triazole bonding, then Cu-BTA complex can be swept away under the mechanical effect of PVA brush scrubbing. Furthermore, the introduction of a nonionic surfactant increases the wettability the Cu-BTA surface, the enhancement of the hydrophilicity can help increase the contact area between copper surface and cleaning solutions, and thus increase the intensity of the reactions occurring on the copper surface. Finally, the removal of BTA residuals could be achieved by the synergetic effect of FA/O chelating agent and nonionic surfactant.

4. Conclusion

Alkaline cleaning solutions consisting of FA/O chelating agents (developed by us) and primary alcohol ethoxylate as nonionic surfactant were formulated to remove adsorbed BTA from the Cu surface. Contact angle and potentiodynamic polarization measurements show that FA/O chelating agent can effectively remove BTA on copper surface. The addition of nonionic surfactant can significantly reduce the surface tension of cleaning solution and facilitate the removal of BTA: as the concentration of nonionic surfactant increased, the capability of FA/O-chelating-agents-based cleaning solution for BTA removal further improved. Total defect counts (TDC) on 300 mm copper patterned wafer dramatically reduced due to the synergetic effect of chelating agent and nonionic surfactant. SEM review shows that chelating agent is efficient for BTA removal, the addition of nonionic surfactant effectively facilitates the removal of BTA.