- Email: [email protected]
Effects of physico-chemical properties between poly(ethyleneimine) and silica abrasive on copper chemical mechanical planarization
Microelectronic Engineering 113 (2014) 50–54
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Effects of physico-chemical properties between poly(ethyleneimine) and silica abrasive on copper chemical mechanical planarization Jihoon Seo a,1, Kwang Seob Yoon b,1, Jinok Moon a,c, Kijung Kim b, Wolfgang Sigmund a,d,⇑, Ungyu Paik a,b,⇑ a
WCU Department of Energy Engineering, Hanyang University, Seoul 133-791, South Korea Department of Nanoscale Semiconductor Engineering, Hanyang University, Seoul 133-791, South Korea c Clean/CMP Technology Team, Memory, Samsung Electronics, Gyeonggi-Do 445-701, South Korea d Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA b
a r t i c l e
i n f o
Article history: Received 9 May 2013 Received in revised form 28 June 2013 Accepted 5 July 2013 Available online 16 July 2013 Keywords: Silica PEI Chemical mechanical planarization Copper Dishing Physico-chemical behavior
a b s t r a c t We investigated the effect of poly (ethyleneimine) (PEI)-modified silica abrasive on the removal rate and the degree of dishing during Cu chemical mechanical planarization (CMP). The PEI-modified silica abrasive was prepared by mutually attractive electrostatic forces between PEI and silica abrasive. The physico-chemical behaviors between PEI and the silica abrasive were evaluated by total organic carbon (TOC), force-separation measurements using atomic force microscopy (AFM) with molecular weight of PEI, which was found to adsorb on silica at pH 7.0 following a Langmuir isotherm. The maximum adsorbed amounts of low and high molecular weight PEI were 0.195 mg/m2 and 0.228 mg/m2, respectively. AFM results showed the repulsive force of the adsorbed PEI layers on the silica surface and the adsorption thickness of PEI on silica vary with the molecular weight of PEI. A twofold change was observed in the AFM analysis. First, the increased areal density of adsorbed PEI caused a higher zeta-potential and longer reaching repulsive force. Second, the adsorption thickness was also significantly enlarged. High molecular weight showed increased adsorption thickness under similar conditions compared to low molecular weight of PEI. These changes of silica abrasive such as electrostatic forces and steric interaction vary with molecular weight of PEI reduced the dishing of Cu pattern film from 50 to 20 nm. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction Chemical mechanical planarization (CMP) for interconnects in ultra large scale integration (ULSI) device has proven as a significant technique for achieving the uniformity of the wafer surface [1]. However, slurry including abrasives induces defects like dishing, erosion and scratches during the Cu CMP process due to the very soft nature of metallic Cu. Controlling the amount of dishing in the damascene process is crucial since it could decrease the conductivity [2,3]. Generally, silica abrasive has been used in Cu CMP, but the direct contact of hard abrasive particles on the Cu inevitably induces defects and dishing in CMP [4,5]. Various efforts have been devoted to overcome the limitations arising due to defects and dishing which includes the integration of Cu electro-polishing with a wetetching technology [6,7], abrasive-free polishing [8–10], and application of core-shell structured abrasive [11–13]. Armini
⇑ Corresponding authors. Tel.: +1 (352) 846 3343; fax: +1 (352) 846 3355 (W. Sigmund), tel.: +82 2 2220 0502; fax: +82 2 2281 0502 (U. Paik). E-mail addresses: [email protected]fl.edu (W. Sigmund), [email protected] (U. Paik). 1 These authors contributed equally to this work. 0167-9317/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.mee.2013.07.006
et al. prepared PMMA core-silica shell abrasives by using coupling agents and electrostatic attraction in order to form the core-shell. In the case of a soft pad in combination with modified abrasives, it shows a remarkable improvement in the defectivity without ant loss in MRR [11]. However, there are few studies in the literature which focus on the effect of the electrostatic forces and steric interaction of polymeric modified silica abrasive on Cu CMP performances. In this study, the effect of polymeric modified silica abrasive on Cu CMP was identified with changes of electrostatic forces and steric interaction,which were controlled by the physico-chemical behavior between polymer and silica abrasive [14]. Many researchers have studied to modify silica surface with polymer such as Poly(ethylene oxide) (PEO) [15,16], Poly(vinyl pyrrolidone) (PVP) [16,17], Polyacrylamide (PAAm) [18], and Poly(ethyleneimine) (PEI) [19–21]. Silica shows a negatively charged surface throughout the pH range 3.0–11 [22]. Cationic polymers have an advantage as modifier over nonionic polymers such as PEO and PVP. In the case of PAAm, very high adsorption amounts of PAAm on silica were observed in literature [18]. Toothick polymer layer on silica causes a decrease of removal rate of Cu. Thus, it is not suitable for CMP application. PEI is a cationic polymer due to the presence of amine groups and is used to change surface potential of negatively charged
51
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
2. Experimental PEI (Sigma-Aldrich Inc.) with two different molecular weights of 2,000 and 25,000, respectively, were added in silica slurry at weight fractions of 0.5% to study the physico-chemical behavior vary with the molecular weight of PEI, which is used for the formation of an adsorption layer on silica abrasive. Commercially available silica slurry (PL-3, FUSO Chemical Co., Tokyo, Japan), which shows particles size with diameter of 35 nm. And 200 mm wafer deposited Cu film was used on the present investigation. Adsorption isotherms for PEI on silica of 35 nm were determined at 25 ± 1 °C via the solution-depletion method using a total organic carbon (TOC) analyzer (5000A, Shimadzu Corp., Kyoto, Japan). Samples were prepared in deionized water containing a silica mass fraction of 5% and equilibrated with rolling for 12 h. The final pH value of the slurry was 7.0. The slurries were centrifuged and the supernatant was further clarified using a 0.02 lm Anotop 25 syringe filter (Whatman, Tokyo, Japan) and then analyzed in triplicate for TOC. TOC values were converted to polymeric molecule mass using an experimentally determined calibration curve. Adsorption isotherms were derived from the difference between the adsorbed amount and the polymer concentration in the supernatant. The electrokinetic behavior of the silica slurry was characterized using the electrokinetic sonic amplitude (ESA) technique (ESA-9800, Matec Applied Sciences, Hopkinton, MA). Separate acid (1.0 N HCl) and base (1.0 N KOH) titrations were performed, beginning at pH 8, and were subsequently combined to generate a complete titration curve. Based on the analysis of the preliminary measurements, the standard uncertainties were ±0.1 mPa m V1 for the ESA measurements and ±0.05 (pH units) for the pH and pHiep values, respectively. To analyze the physico-chemical behavior of the PEI on the silicon oxide film, force-separation measurements were obtained using a commercial multimode atomic force microscope (XE-150, Park System, Korea). The Si atomic force microscopy (AFM) tip (SICONGG, AppNano) was used to investigate the interaction forces using contact mode. The spring constant of the cantilever provided by the manufacturer was 0.12 N/m. Also, SiO2 film was prepared as the substrate with a size of 20 mm 20 mm and immersed SiO2 substrate in PEI solution for 20 min. Then solution with the same pH was used to remove the side effect by non-adsorbed polymer molecules. The concentration of the PEI solution was 0.5 wt%, and the final pH of PEI and washing solutions prepared in this study was 7.0. The force displacement curves were converted from deflection signal-piezo scanner movement data and averaged using data analysis software (XEI, Park System). The experimental procedure used in this study was the same as reported in our previous studies [23].
Table 1 Experimental conditions. Slurry
Silica slurry
Solid concentration of slurry Pad Table speed Down force Flow rate
5.0 wt% IC1000 SUVAIV 80 rpm 20.68 kPa 100 ml/min
The CMP field evaluation was performed using three kinds of slurries with PEI. A Cu coupon wafer (6 cm 6 cm) CMP tool (POLI-300, G&P, Korea) was used. The polishing pad was a grooved IC1000/SubaIV (Rodel). The Cu film thickness was measured using a four point probe (CMT-SR2000N, Chang Min Tech, Korea) to calculate removal rates. The polishing test conditions are shown in Table 1. To identify the surface properties such as microscratches and defects of Cu film after CMP, the optical images were taken by light microscopy (BX51, Olympus, Tokyo, Japan). 3. Results and discussion Fig. 1 shows the adsorption isotherm of PEI on silica abrasive as function of molecular weight at pH 7.0. The experimental data was modeled using the linearized Langmuir adsorption isotherm equation.
C e =Q m þ 1=bQ m ¼ C e =Q e Qe is the adsorbed amount of PEI per silica surface area at equilibrium (mg/m2), Ce is PEI concentration in the bulk solution (mg/ L), Qm is the maximum adsorbed amount (mg/m2), and b is the free energy constant of adsorption (L/ mg). The silica slurry has a negative charge above pH 3.5, which is the isoelectric point of silica, and PEI has a positive charge since pKa of PEI is 10.5. Consequently, PEI is adsorbed on the silica most likely governed by electrostatic force. The adsorption follows a Langmuir isotherm that is as the concentration of PEI increases the adsorbed amount of PEI increases until it reaches a plateau region. The adsorbed amount of PEI 2,000 to (<350 mg/L) was almost 100%, followed by a flattening and reached a plateau level of 0.195 mg/m2. In PEI 25,000, the adsorbed amount increased sharply at a low residual concentration and reached a plateau level of 0.228 mg/m2. These results show PEI 25,000 has a higher affinity isotherm compared to PEI 2,000. We hypothesize that there are two factors contributing to a higher
0.25
Adsorption admount of PEI 2 on Silica (mg/m )
surface of silica abrasive [11,19–21]. This PEI adsorption layer prevents the silica abrasive from the direct contact on Cu [13]. Although PEI-modified silica shows a decreased the blanket removal rate of Cu film due to a decrease of the hardness of silica abrasive by PEI adsorption layer on silica abrasive, it shows the improved polishing performances of pattern wafer such as dishing. The steric interaction of PEI adsorption layer on silica abrasive induces to decrease the degree of dishing due to the low friction force by a decrease of the hardness of silica abrasive. Therefore, the polishing performances of pattern waferwere improved with slurry containing PEI-modified silica abrasive. As a control the physico-chemical behavior between PEI and silica abrasive vary with the molecular weight of PEI, the polishing performances of Cu CMP were identified. Also, PEI-modified silica abrasive was effective for preventing the formation of microscratches during Cu CMP.
0.20 0.15 0.10 0.05 PEI Mw 2000 PEI Mw 25000
0.00 0
200
400
600
800
1000
1200
PEI concentration (mg/L) Fig. 1. Room temperature adsorption isotherms of PEI on silica abrasive at pH 7. Data was fitted with the Langmuir adsorption isotherm equation.
52
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
20
20 PEI Mw 2000 PEI Mw 25000
0
Step height (nm)
10
10
PEI Mw 2000 PEI Mw 25000
10
Force (nN)
Force (nN)
15
1
5 0
10
20
30
40
50
60
70
80
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
-10 -20 -30
90 100
-40
Separation (nm)
0
-50
0
10
20
30
40
50
60
70
80
90
0
100
20
Separation (nm)
40
60
Line width (50um) (um)
Fig. 2. Force–distance curves between AFM tip and oxide film as a function of molecular weight of PEI. Solid line denotes the Debye length.
20
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
10
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
ESA (mPa*M/V)
0.2 0.1 0.0
Step height (nm)
0 0.3
-10 -20 -30 -40
-0.1
-50
-0.2
0
20
40
60
80
100
Line width (100um) (um) -0.3 Fig. 5. Results of the CMP evaluation as a function of the molecular weight of PEI: (a) 50 lm line width and (b) 100 lm line width.
-0.4 4
5
6
7
8
9
10
11
pH Fig. 3. Electrokinetic behaviors of silica slurry and silica slurry with PEI added as a function of slurry pH.
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
Removal rate (nm/min)
400
300
200
100
0 -3
-2
-1
0
1
2
3
Position (cm) Fig. 4. Removal rate of Cu film as a function of the molecular weight of PEI.
adsorbed amount. First the high molecular weight compound is made via cross-linking of shorter molecules. This means that three types of amine groups are present. Especially tertiary amine groups have a higher pKa value and are therefore expected to drive the increased adsorption density. Furthermore, the increase in molecular weight will also cause an increase in adsorbed amount per area due to the expansion into the liquid. The force displacement curves between the Si AFM tip and the SiO2 film were measured to identify the physico-chemical behavior between PEI and SiO2 film as a function of molecular weight of PEI. Fig. 2 shows that two distinct sections can be seen. First, both regions have linear behavior in a semi-logarithmic plot. This indicates that the long range forces are derived solely from electrical double layer interactions following DLVO theory [24]. As the AFM tip approaches the substrate, a second repulsive curve with steeper slope occurs, yet still showing a linear behavior. This steeper slope is assigned to steric repulsion of the PEI layers adsorbed on the SiO2 film and AFM tip. In reality, these forces are caused by an electrostatic layer compression since both adsorbed layersare highly positively charged. The sensitivity of the AFM limits the detection of repulsive forces in PEI 25,000 and PEI 2,000 solution at pH 7.0 at a separation of 65 nm and 50 nm, respectively. The different parallel lines above 5 and 10 nm is electrostatic potential effect of DLVO theory and reveal different electrostatic potential which is generated PEI 2,000 and PEI 25,000 adsorbed on silica abrasive. This
53
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
(a)
X 50
(b)
5 µm
25 mm
(c)
X 50
X 200
(d)
25 mm
X 200
5 µm
Fig. 6. Optical images of Cu film after CMP. Silica slurry at magnifications of (a) 50, (b) 200. Silica slurry with PEI at magnifications of (a) 50, (b) 200.
result could be also confirmed in the logarithmic plot in Fig. 2. The steric potential which concerns entropy hindrance and steric stabilization take over the interaction below 5 and 10 nm that indicate the steric effect of PEI layer on the SiO2 substrate is 2.5 nm for PEI 2,000 and 5 nm for PEI 25,000. Electrokinetic behaviors of silica and modified silica as a function of PEI molecular weight were investigated as shown in Fig. 3. The ESA values of silica with PEI 2,000 and 25,000 at pH 7 are 0.165 and 0.200 mPa m V1, respectively. The electrokinetic behavior reflects in the interaction between the particles in CMP slurry. The ESA values of silica were negative above pH 3.5 [25], which is the isoelectric point of silica. With PEI, the ESA values of particle were changed. The isoelectric point is shifted from pH 3.5 to 10.0 since the presence of polymer chains disturbs the hydrodynamic plane of shear, shifting it further out from the particle surface. The surface charge of particle is changed from negative to positive due to PEI which is cationicpolymer. The removal rates for Cu film are shown in Fig. 4. The removal performance was evaluated without and varies with molecular weight of PEI. In case of silica slurry without PEI, the average removal rate of Cu film was 300 nm/min. the silica abrasive scrub the wafer surface strongly with high friction forces due to the direct contact of hard abrasive particles on the Cu. When silica abrasive with PEI was modified in slurry, the removal rate was changed. The average removal rates of the Cu film were 240 and 180 nm/min for PEI with Mw 2,000 and Mw 25,000, respectively. As the molecular weight of PEI increase, the removal rate of Cu film was decreased due to the low friction force by a decrease of the hardness of silica abrasive. Although the silica adsorbed PEI 25,000 at pH 7 has thicker PEI layer and higher electrical potential than adsorbed PEI 2,000 it results in decreasing the removal rate. The characteristics of the CMP evaluation are attributed to the modified silica abrasive. The removal rate of the slurry without PEI abrasive is higher for the Cu film due to mechanical effects than that of the modified silica abrasive by PEI.
Fig. 5 shows the information of the line profiles of the 50 lm and 100 lm wide pattern trenches which were measured by atomic force microscopy (AFM). The line profile showed dishing results of polished Cu pattern in the CMP process as function of molecular weight. The dishing amounts of silica slurry with PEI was lower than those of silica slurry in 50 lm and 100 lm patterns. Polishing with PEI 2,000 and 25,000 decreased the dishing to reach a final value of 20 nm and 30 nm respectively. The surface properties such as microscratchs of Cu film after CMP are shown in Fig. 6. The microscratchs and defects on Cu film decreased with PEI-modification of silica abrasive. It can be explained the reason why the steric interaction of PEI adsorption layer on silica abrasive improve the surface properties due to the low friction force by a decrease of the hardness of silica abrasive. The removal rate of blanket Cu wafer was decreased by a decrease of the hardness of PEI modified silica abrasive. However, the CMP process of pattern Cu wafer with PEI modified silica abrasive shows a improved the degree of dishing due to low friction force by a decrease of the hardness of abrasive. The characteristics of the CMP evaluation such as surface roughness, the degree of dishing, and defect are more important than removal rate of blanket wafer. Consequently, the surface modification of abrasive in polishing presumably could control dishing to get better results in CMP process.
4. Conclusion We investigated the physico-chemical behavior of modified silica abrasive as function of molecular weight of PEI in Cu CMP. The electrostatic force and steric interaction of adsorbed PEI on silica abrasive were changed by molecular weight of PEI. The adsorbed amount of PEI on the silica particle increased from 0.195 to 0.228 mg/m2 as the molecular weight increases from 2,000 to 25,000 at pH 7.0. The steric repulsion of PEI on the SiO2 film gets
54
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
dominated by electric double layer repulsion at 5 nm for PAA 2,000 and 10 nm for PAA 25,000. This steric effect is hypothesized to cause a reduction in friction and number of mechanically damaging penetrations of abrasive particles into the Cu film. As for the adsorption of PEI on silica abrasive, the isoelectric point of silica slurry is shifted from pH 3.5–10 and the surface charge of particles is changed at pH 7.0. The CMP performancewas evaluated with and without PEI and dishing results with PEI 2,000 and 25,000 indicated decreased dishing with final values of 30 nm and 20 nm, respectively. Also, PEI-modified silica abrasive improves the surface properties of Cu film and it is effective for preventing the formation of microscratches during Cu CMP. We conclude that surface modifications of abrasive particles using adsorbed polymers with strong bonding may be a critical factor in reducing the Cu dishing and improving the surface properties. Acknowledgments This work was supported by the Global Research Laboratory (GRL) Program (K20704000003TA050000310) through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT (Information and Communication Technologies) and Future Planning, and the World Class University (WCU) Program (R31-10092) through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT (Information and Communication Technologies) and Future Planning. References [1] X.D. Feng, D.C. Sayle, Z.L. Wang, M.S. Paras, B. Santora, A.C. Sutorik, T.X.T. Sayle, Y. Yang, Y. Ding, X.D. Wang, Y.S. Her, Science 312 (2006) 1504–1508.
[2] C.C. Hung, W.H. Lee, Y.S. Wang, Y.R. Chen, Jpn. J. Appl. Phys. 47 (2008) 989– 991. [3] R.Z. Chang, Y. Cao, C.J. Spanos, IEEE T. Electron Dev. 51 (2004) 1577–1583. [4] Y. Ein-Eli, D. Starosvetsky, Electrochim. Acta 52 (2007) 1825–1838. [5] Y. Ahn, J.H. Yoon, C.W. Baek, Y.K. Kim, Wear 257 (2004) 785–789. [6] P. Cojocaru, F. Muscolino, L. Magagnin, Microelectron. Eng. 87 (2010) 2187– 2189. [7] S.C. Chang, J.M. Shieh, C.C. Huang, B.T. Dai, M.S. Feng, Jpn. J. Appl. Phys. 1 (41) (2002) 7332–7337. [8] T. Haque, S. Balakumar, A.S. Kumar, M. Rahman, J. Electrochem. Soc. 152 (2005) G417–G422. [9] J.Y. Fang, M.S. Tsai, B.T. Dai, Y.S. Wu, M.S. Feng, J. Electrochem. Soc. 153 (2006) G44–G46. [10] S. Ramakrishnan, S.V.S.B. Janjam, U.B. Patri, D. Roy, S.V. Babu, Microelectron. Eng. 84 (2007) 80–86. [11] S. Armini, C.M. Whelan, M. Moinpour, K. Maex, J. Electrochem. Soc. 156 (2009) H18–H26. [12] Y. Chen, J.X. Lu, Z.G. Chen, Microelectron. Eng. 88 (2011) 200–205. [13] W.B. Mu, M. Fu, Microelectron. Eng. 96 (2012) 51–55. [14] Y.H. Kim, S.M. Lee, K.J. Lee, U. Paik, J.G. Park, J. Mater. Res. 23 (2008) 49–54. [15] J. Rubio, J. Kitchener, J. Colloid Interface Sci. 57 (1976) 132–142. [16] R.S. Parnas, M. Chaimberg, V. Taepaisitphongse, Y. Cohen, J. Colloid Interface Sci. 129 (1989) 441–450. [17] C. Stuart, G. Fleer, B. Bijsterbosch, J. Colloid Interface Sci. 90 (1982) 310–320. [18] V. Shubin, J. Colloid Interface Sci. 191 (1997) 372–377. [19] R. Meszaros, L. Thompson, M. Bos, P. de Groot, Langmuir 18 (2002) 6164–6169. [20] R. Meszaros, I. Varga, T. Gilanyi, Langmuir 20 (2004) 5026–5029. [21] G.M. Lindquist, R.A. Stratton, J. Colloid Interface Sci. 55 (1976) 45–59. [22] W. Choi, S.-M. Lee, R.K. Singh, Electrochem. Solid-State Lett. 7 (2004) G141– G144. [23] C.W. Cho, S.K. Kim, U. Paik, J.G. Park, W.M. Sigmund, J. Mater. Res. 21 (2006) 473–479. [24] P. Wrschka, J. Hernandez, G.S. Oehrlein, J.A. Negrych, G. Haag, P. Rau, J.E. Currie, J. Electrochem. Soc. 148 (2001) G321–G325. [25] V.A. Hackley, U. Paik, B.H. Kim, S.G. Malghan, J. Am. Ceram. Soc. 80 (1997) 1781–1788.
Contents lists available at ScienceDirect
Microelectronic Engineering journal homepage: www.elsevier.com/locate/mee
Effects of physico-chemical properties between poly(ethyleneimine) and silica abrasive on copper chemical mechanical planarization Jihoon Seo a,1, Kwang Seob Yoon b,1, Jinok Moon a,c, Kijung Kim b, Wolfgang Sigmund a,d,⇑, Ungyu Paik a,b,⇑ a
WCU Department of Energy Engineering, Hanyang University, Seoul 133-791, South Korea Department of Nanoscale Semiconductor Engineering, Hanyang University, Seoul 133-791, South Korea c Clean/CMP Technology Team, Memory, Samsung Electronics, Gyeonggi-Do 445-701, South Korea d Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA b
a r t i c l e
i n f o
Article history: Received 9 May 2013 Received in revised form 28 June 2013 Accepted 5 July 2013 Available online 16 July 2013 Keywords: Silica PEI Chemical mechanical planarization Copper Dishing Physico-chemical behavior
a b s t r a c t We investigated the effect of poly (ethyleneimine) (PEI)-modified silica abrasive on the removal rate and the degree of dishing during Cu chemical mechanical planarization (CMP). The PEI-modified silica abrasive was prepared by mutually attractive electrostatic forces between PEI and silica abrasive. The physico-chemical behaviors between PEI and the silica abrasive were evaluated by total organic carbon (TOC), force-separation measurements using atomic force microscopy (AFM) with molecular weight of PEI, which was found to adsorb on silica at pH 7.0 following a Langmuir isotherm. The maximum adsorbed amounts of low and high molecular weight PEI were 0.195 mg/m2 and 0.228 mg/m2, respectively. AFM results showed the repulsive force of the adsorbed PEI layers on the silica surface and the adsorption thickness of PEI on silica vary with the molecular weight of PEI. A twofold change was observed in the AFM analysis. First, the increased areal density of adsorbed PEI caused a higher zeta-potential and longer reaching repulsive force. Second, the adsorption thickness was also significantly enlarged. High molecular weight showed increased adsorption thickness under similar conditions compared to low molecular weight of PEI. These changes of silica abrasive such as electrostatic forces and steric interaction vary with molecular weight of PEI reduced the dishing of Cu pattern film from 50 to 20 nm. Ó 2013 Elsevier B.V. All rights reserved.
1. Introduction Chemical mechanical planarization (CMP) for interconnects in ultra large scale integration (ULSI) device has proven as a significant technique for achieving the uniformity of the wafer surface [1]. However, slurry including abrasives induces defects like dishing, erosion and scratches during the Cu CMP process due to the very soft nature of metallic Cu. Controlling the amount of dishing in the damascene process is crucial since it could decrease the conductivity [2,3]. Generally, silica abrasive has been used in Cu CMP, but the direct contact of hard abrasive particles on the Cu inevitably induces defects and dishing in CMP [4,5]. Various efforts have been devoted to overcome the limitations arising due to defects and dishing which includes the integration of Cu electro-polishing with a wetetching technology [6,7], abrasive-free polishing [8–10], and application of core-shell structured abrasive [11–13]. Armini
⇑ Corresponding authors. Tel.: +1 (352) 846 3343; fax: +1 (352) 846 3355 (W. Sigmund), tel.: +82 2 2220 0502; fax: +82 2 2281 0502 (U. Paik). E-mail addresses: [email protected]fl.edu (W. Sigmund), [email protected] (U. Paik). 1 These authors contributed equally to this work. 0167-9317/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.mee.2013.07.006
et al. prepared PMMA core-silica shell abrasives by using coupling agents and electrostatic attraction in order to form the core-shell. In the case of a soft pad in combination with modified abrasives, it shows a remarkable improvement in the defectivity without ant loss in MRR [11]. However, there are few studies in the literature which focus on the effect of the electrostatic forces and steric interaction of polymeric modified silica abrasive on Cu CMP performances. In this study, the effect of polymeric modified silica abrasive on Cu CMP was identified with changes of electrostatic forces and steric interaction,which were controlled by the physico-chemical behavior between polymer and silica abrasive [14]. Many researchers have studied to modify silica surface with polymer such as Poly(ethylene oxide) (PEO) [15,16], Poly(vinyl pyrrolidone) (PVP) [16,17], Polyacrylamide (PAAm) [18], and Poly(ethyleneimine) (PEI) [19–21]. Silica shows a negatively charged surface throughout the pH range 3.0–11 [22]. Cationic polymers have an advantage as modifier over nonionic polymers such as PEO and PVP. In the case of PAAm, very high adsorption amounts of PAAm on silica were observed in literature [18]. Toothick polymer layer on silica causes a decrease of removal rate of Cu. Thus, it is not suitable for CMP application. PEI is a cationic polymer due to the presence of amine groups and is used to change surface potential of negatively charged
51
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
2. Experimental PEI (Sigma-Aldrich Inc.) with two different molecular weights of 2,000 and 25,000, respectively, were added in silica slurry at weight fractions of 0.5% to study the physico-chemical behavior vary with the molecular weight of PEI, which is used for the formation of an adsorption layer on silica abrasive. Commercially available silica slurry (PL-3, FUSO Chemical Co., Tokyo, Japan), which shows particles size with diameter of 35 nm. And 200 mm wafer deposited Cu film was used on the present investigation. Adsorption isotherms for PEI on silica of 35 nm were determined at 25 ± 1 °C via the solution-depletion method using a total organic carbon (TOC) analyzer (5000A, Shimadzu Corp., Kyoto, Japan). Samples were prepared in deionized water containing a silica mass fraction of 5% and equilibrated with rolling for 12 h. The final pH value of the slurry was 7.0. The slurries were centrifuged and the supernatant was further clarified using a 0.02 lm Anotop 25 syringe filter (Whatman, Tokyo, Japan) and then analyzed in triplicate for TOC. TOC values were converted to polymeric molecule mass using an experimentally determined calibration curve. Adsorption isotherms were derived from the difference between the adsorbed amount and the polymer concentration in the supernatant. The electrokinetic behavior of the silica slurry was characterized using the electrokinetic sonic amplitude (ESA) technique (ESA-9800, Matec Applied Sciences, Hopkinton, MA). Separate acid (1.0 N HCl) and base (1.0 N KOH) titrations were performed, beginning at pH 8, and were subsequently combined to generate a complete titration curve. Based on the analysis of the preliminary measurements, the standard uncertainties were ±0.1 mPa m V1 for the ESA measurements and ±0.05 (pH units) for the pH and pHiep values, respectively. To analyze the physico-chemical behavior of the PEI on the silicon oxide film, force-separation measurements were obtained using a commercial multimode atomic force microscope (XE-150, Park System, Korea). The Si atomic force microscopy (AFM) tip (SICONGG, AppNano) was used to investigate the interaction forces using contact mode. The spring constant of the cantilever provided by the manufacturer was 0.12 N/m. Also, SiO2 film was prepared as the substrate with a size of 20 mm 20 mm and immersed SiO2 substrate in PEI solution for 20 min. Then solution with the same pH was used to remove the side effect by non-adsorbed polymer molecules. The concentration of the PEI solution was 0.5 wt%, and the final pH of PEI and washing solutions prepared in this study was 7.0. The force displacement curves were converted from deflection signal-piezo scanner movement data and averaged using data analysis software (XEI, Park System). The experimental procedure used in this study was the same as reported in our previous studies [23].
Table 1 Experimental conditions. Slurry
Silica slurry
Solid concentration of slurry Pad Table speed Down force Flow rate
5.0 wt% IC1000 SUVAIV 80 rpm 20.68 kPa 100 ml/min
The CMP field evaluation was performed using three kinds of slurries with PEI. A Cu coupon wafer (6 cm 6 cm) CMP tool (POLI-300, G&P, Korea) was used. The polishing pad was a grooved IC1000/SubaIV (Rodel). The Cu film thickness was measured using a four point probe (CMT-SR2000N, Chang Min Tech, Korea) to calculate removal rates. The polishing test conditions are shown in Table 1. To identify the surface properties such as microscratches and defects of Cu film after CMP, the optical images were taken by light microscopy (BX51, Olympus, Tokyo, Japan). 3. Results and discussion Fig. 1 shows the adsorption isotherm of PEI on silica abrasive as function of molecular weight at pH 7.0. The experimental data was modeled using the linearized Langmuir adsorption isotherm equation.
C e =Q m þ 1=bQ m ¼ C e =Q e Qe is the adsorbed amount of PEI per silica surface area at equilibrium (mg/m2), Ce is PEI concentration in the bulk solution (mg/ L), Qm is the maximum adsorbed amount (mg/m2), and b is the free energy constant of adsorption (L/ mg). The silica slurry has a negative charge above pH 3.5, which is the isoelectric point of silica, and PEI has a positive charge since pKa of PEI is 10.5. Consequently, PEI is adsorbed on the silica most likely governed by electrostatic force. The adsorption follows a Langmuir isotherm that is as the concentration of PEI increases the adsorbed amount of PEI increases until it reaches a plateau region. The adsorbed amount of PEI 2,000 to (<350 mg/L) was almost 100%, followed by a flattening and reached a plateau level of 0.195 mg/m2. In PEI 25,000, the adsorbed amount increased sharply at a low residual concentration and reached a plateau level of 0.228 mg/m2. These results show PEI 25,000 has a higher affinity isotherm compared to PEI 2,000. We hypothesize that there are two factors contributing to a higher
0.25
Adsorption admount of PEI 2 on Silica (mg/m )
surface of silica abrasive [11,19–21]. This PEI adsorption layer prevents the silica abrasive from the direct contact on Cu [13]. Although PEI-modified silica shows a decreased the blanket removal rate of Cu film due to a decrease of the hardness of silica abrasive by PEI adsorption layer on silica abrasive, it shows the improved polishing performances of pattern wafer such as dishing. The steric interaction of PEI adsorption layer on silica abrasive induces to decrease the degree of dishing due to the low friction force by a decrease of the hardness of silica abrasive. Therefore, the polishing performances of pattern waferwere improved with slurry containing PEI-modified silica abrasive. As a control the physico-chemical behavior between PEI and silica abrasive vary with the molecular weight of PEI, the polishing performances of Cu CMP were identified. Also, PEI-modified silica abrasive was effective for preventing the formation of microscratches during Cu CMP.
0.20 0.15 0.10 0.05 PEI Mw 2000 PEI Mw 25000
0.00 0
200
400
600
800
1000
1200
PEI concentration (mg/L) Fig. 1. Room temperature adsorption isotherms of PEI on silica abrasive at pH 7. Data was fitted with the Langmuir adsorption isotherm equation.
52
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
20
20 PEI Mw 2000 PEI Mw 25000
0
Step height (nm)
10
10
PEI Mw 2000 PEI Mw 25000
10
Force (nN)
Force (nN)
15
1
5 0
10
20
30
40
50
60
70
80
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
-10 -20 -30
90 100
-40
Separation (nm)
0
-50
0
10
20
30
40
50
60
70
80
90
0
100
20
Separation (nm)
40
60
Line width (50um) (um)
Fig. 2. Force–distance curves between AFM tip and oxide film as a function of molecular weight of PEI. Solid line denotes the Debye length.
20
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
10
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
ESA (mPa*M/V)
0.2 0.1 0.0
Step height (nm)
0 0.3
-10 -20 -30 -40
-0.1
-50
-0.2
0
20
40
60
80
100
Line width (100um) (um) -0.3 Fig. 5. Results of the CMP evaluation as a function of the molecular weight of PEI: (a) 50 lm line width and (b) 100 lm line width.
-0.4 4
5
6
7
8
9
10
11
pH Fig. 3. Electrokinetic behaviors of silica slurry and silica slurry with PEI added as a function of slurry pH.
Silica Silica + PEI Mw 2000 Silica + PEI Mw 25000
Removal rate (nm/min)
400
300
200
100
0 -3
-2
-1
0
1
2
3
Position (cm) Fig. 4. Removal rate of Cu film as a function of the molecular weight of PEI.
adsorbed amount. First the high molecular weight compound is made via cross-linking of shorter molecules. This means that three types of amine groups are present. Especially tertiary amine groups have a higher pKa value and are therefore expected to drive the increased adsorption density. Furthermore, the increase in molecular weight will also cause an increase in adsorbed amount per area due to the expansion into the liquid. The force displacement curves between the Si AFM tip and the SiO2 film were measured to identify the physico-chemical behavior between PEI and SiO2 film as a function of molecular weight of PEI. Fig. 2 shows that two distinct sections can be seen. First, both regions have linear behavior in a semi-logarithmic plot. This indicates that the long range forces are derived solely from electrical double layer interactions following DLVO theory [24]. As the AFM tip approaches the substrate, a second repulsive curve with steeper slope occurs, yet still showing a linear behavior. This steeper slope is assigned to steric repulsion of the PEI layers adsorbed on the SiO2 film and AFM tip. In reality, these forces are caused by an electrostatic layer compression since both adsorbed layersare highly positively charged. The sensitivity of the AFM limits the detection of repulsive forces in PEI 25,000 and PEI 2,000 solution at pH 7.0 at a separation of 65 nm and 50 nm, respectively. The different parallel lines above 5 and 10 nm is electrostatic potential effect of DLVO theory and reveal different electrostatic potential which is generated PEI 2,000 and PEI 25,000 adsorbed on silica abrasive. This
53
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
(a)
X 50
(b)
5 µm
25 mm
(c)
X 50
X 200
(d)
25 mm
X 200
5 µm
Fig. 6. Optical images of Cu film after CMP. Silica slurry at magnifications of (a) 50, (b) 200. Silica slurry with PEI at magnifications of (a) 50, (b) 200.
result could be also confirmed in the logarithmic plot in Fig. 2. The steric potential which concerns entropy hindrance and steric stabilization take over the interaction below 5 and 10 nm that indicate the steric effect of PEI layer on the SiO2 substrate is 2.5 nm for PEI 2,000 and 5 nm for PEI 25,000. Electrokinetic behaviors of silica and modified silica as a function of PEI molecular weight were investigated as shown in Fig. 3. The ESA values of silica with PEI 2,000 and 25,000 at pH 7 are 0.165 and 0.200 mPa m V1, respectively. The electrokinetic behavior reflects in the interaction between the particles in CMP slurry. The ESA values of silica were negative above pH 3.5 [25], which is the isoelectric point of silica. With PEI, the ESA values of particle were changed. The isoelectric point is shifted from pH 3.5 to 10.0 since the presence of polymer chains disturbs the hydrodynamic plane of shear, shifting it further out from the particle surface. The surface charge of particle is changed from negative to positive due to PEI which is cationicpolymer. The removal rates for Cu film are shown in Fig. 4. The removal performance was evaluated without and varies with molecular weight of PEI. In case of silica slurry without PEI, the average removal rate of Cu film was 300 nm/min. the silica abrasive scrub the wafer surface strongly with high friction forces due to the direct contact of hard abrasive particles on the Cu. When silica abrasive with PEI was modified in slurry, the removal rate was changed. The average removal rates of the Cu film were 240 and 180 nm/min for PEI with Mw 2,000 and Mw 25,000, respectively. As the molecular weight of PEI increase, the removal rate of Cu film was decreased due to the low friction force by a decrease of the hardness of silica abrasive. Although the silica adsorbed PEI 25,000 at pH 7 has thicker PEI layer and higher electrical potential than adsorbed PEI 2,000 it results in decreasing the removal rate. The characteristics of the CMP evaluation are attributed to the modified silica abrasive. The removal rate of the slurry without PEI abrasive is higher for the Cu film due to mechanical effects than that of the modified silica abrasive by PEI.
Fig. 5 shows the information of the line profiles of the 50 lm and 100 lm wide pattern trenches which were measured by atomic force microscopy (AFM). The line profile showed dishing results of polished Cu pattern in the CMP process as function of molecular weight. The dishing amounts of silica slurry with PEI was lower than those of silica slurry in 50 lm and 100 lm patterns. Polishing with PEI 2,000 and 25,000 decreased the dishing to reach a final value of 20 nm and 30 nm respectively. The surface properties such as microscratchs of Cu film after CMP are shown in Fig. 6. The microscratchs and defects on Cu film decreased with PEI-modification of silica abrasive. It can be explained the reason why the steric interaction of PEI adsorption layer on silica abrasive improve the surface properties due to the low friction force by a decrease of the hardness of silica abrasive. The removal rate of blanket Cu wafer was decreased by a decrease of the hardness of PEI modified silica abrasive. However, the CMP process of pattern Cu wafer with PEI modified silica abrasive shows a improved the degree of dishing due to low friction force by a decrease of the hardness of abrasive. The characteristics of the CMP evaluation such as surface roughness, the degree of dishing, and defect are more important than removal rate of blanket wafer. Consequently, the surface modification of abrasive in polishing presumably could control dishing to get better results in CMP process.
4. Conclusion We investigated the physico-chemical behavior of modified silica abrasive as function of molecular weight of PEI in Cu CMP. The electrostatic force and steric interaction of adsorbed PEI on silica abrasive were changed by molecular weight of PEI. The adsorbed amount of PEI on the silica particle increased from 0.195 to 0.228 mg/m2 as the molecular weight increases from 2,000 to 25,000 at pH 7.0. The steric repulsion of PEI on the SiO2 film gets
54
J. Seo et al. / Microelectronic Engineering 113 (2014) 50–54
dominated by electric double layer repulsion at 5 nm for PAA 2,000 and 10 nm for PAA 25,000. This steric effect is hypothesized to cause a reduction in friction and number of mechanically damaging penetrations of abrasive particles into the Cu film. As for the adsorption of PEI on silica abrasive, the isoelectric point of silica slurry is shifted from pH 3.5–10 and the surface charge of particles is changed at pH 7.0. The CMP performancewas evaluated with and without PEI and dishing results with PEI 2,000 and 25,000 indicated decreased dishing with final values of 30 nm and 20 nm, respectively. Also, PEI-modified silica abrasive improves the surface properties of Cu film and it is effective for preventing the formation of microscratches during Cu CMP. We conclude that surface modifications of abrasive particles using adsorbed polymers with strong bonding may be a critical factor in reducing the Cu dishing and improving the surface properties. Acknowledgments This work was supported by the Global Research Laboratory (GRL) Program (K20704000003TA050000310) through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT (Information and Communication Technologies) and Future Planning, and the World Class University (WCU) Program (R31-10092) through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT (Information and Communication Technologies) and Future Planning. References [1] X.D. Feng, D.C. Sayle, Z.L. Wang, M.S. Paras, B. Santora, A.C. Sutorik, T.X.T. Sayle, Y. Yang, Y. Ding, X.D. Wang, Y.S. Her, Science 312 (2006) 1504–1508.
[2] C.C. Hung, W.H. Lee, Y.S. Wang, Y.R. Chen, Jpn. J. Appl. Phys. 47 (2008) 989– 991. [3] R.Z. Chang, Y. Cao, C.J. Spanos, IEEE T. Electron Dev. 51 (2004) 1577–1583. [4] Y. Ein-Eli, D. Starosvetsky, Electrochim. Acta 52 (2007) 1825–1838. [5] Y. Ahn, J.H. Yoon, C.W. Baek, Y.K. Kim, Wear 257 (2004) 785–789. [6] P. Cojocaru, F. Muscolino, L. Magagnin, Microelectron. Eng. 87 (2010) 2187– 2189. [7] S.C. Chang, J.M. Shieh, C.C. Huang, B.T. Dai, M.S. Feng, Jpn. J. Appl. Phys. 1 (41) (2002) 7332–7337. [8] T. Haque, S. Balakumar, A.S. Kumar, M. Rahman, J. Electrochem. Soc. 152 (2005) G417–G422. [9] J.Y. Fang, M.S. Tsai, B.T. Dai, Y.S. Wu, M.S. Feng, J. Electrochem. Soc. 153 (2006) G44–G46. [10] S. Ramakrishnan, S.V.S.B. Janjam, U.B. Patri, D. Roy, S.V. Babu, Microelectron. Eng. 84 (2007) 80–86. [11] S. Armini, C.M. Whelan, M. Moinpour, K. Maex, J. Electrochem. Soc. 156 (2009) H18–H26. [12] Y. Chen, J.X. Lu, Z.G. Chen, Microelectron. Eng. 88 (2011) 200–205. [13] W.B. Mu, M. Fu, Microelectron. Eng. 96 (2012) 51–55. [14] Y.H. Kim, S.M. Lee, K.J. Lee, U. Paik, J.G. Park, J. Mater. Res. 23 (2008) 49–54. [15] J. Rubio, J. Kitchener, J. Colloid Interface Sci. 57 (1976) 132–142. [16] R.S. Parnas, M. Chaimberg, V. Taepaisitphongse, Y. Cohen, J. Colloid Interface Sci. 129 (1989) 441–450. [17] C. Stuart, G. Fleer, B. Bijsterbosch, J. Colloid Interface Sci. 90 (1982) 310–320. [18] V. Shubin, J. Colloid Interface Sci. 191 (1997) 372–377. [19] R. Meszaros, L. Thompson, M. Bos, P. de Groot, Langmuir 18 (2002) 6164–6169. [20] R. Meszaros, I. Varga, T. Gilanyi, Langmuir 20 (2004) 5026–5029. [21] G.M. Lindquist, R.A. Stratton, J. Colloid Interface Sci. 55 (1976) 45–59. [22] W. Choi, S.-M. Lee, R.K. Singh, Electrochem. Solid-State Lett. 7 (2004) G141– G144. [23] C.W. Cho, S.K. Kim, U. Paik, J.G. Park, W.M. Sigmund, J. Mater. Res. 21 (2006) 473–479. [24] P. Wrschka, J. Hernandez, G.S. Oehrlein, J.A. Negrych, G. Haag, P. Rau, J.E. Currie, J. Electrochem. Soc. 148 (2001) G321–G325. [25] V.A. Hackley, U. Paik, B.H. Kim, S.G. Malghan, J. Am. Ceram. Soc. 80 (1997) 1781–1788.