Glycolic acid in hydrogen peroxide-based slurry for enhancing copper chemical mechanical polishing

(1)

Glycolic acid in hydrogen peroxide-based slurry for

enhancing copper chemical mechanical polishing

Tzu-Hsuan Tsai

a,*

, Yung-Fu Wu

b

, Shi-Chern Yen

b

a

Department of Chemical Engineering, Northern Taiwan Institute of Science and Technology, No. 2, Xueyuan Rd., Beitou, Taipei 112, Taiwan, R.O.C.

b_{Department of Chemical Engineering, National Taiwan University, Taipei 106, Taiwan} Received 14 August 2004; accepted 30 October 2004

Available online 19 November 2004

Abstract

The effects of glycolic acid (GCA) added into hydrogen peroxide (H2O2) or urea-hydrogen peroxide (U-H2O2) slurries on Cu-CMP performance were investigated. Experiments showed that GCA could prevent H2O2 or U-H2O2from rapid decomposition and increase the active peroxide lifetime of the slurries. In addition, electrochemical studies from polarization and impedance experiments verified that copper removal efficiency could be enhanced by use of GCA. Meanwhile, a valid equivalent circuit for Cu-CMP system was proposed, and the fitting results provided a good index to surface planarization. Furthermore, GCA could shorten the range of isoelectric points between Cu film and a-Al2O3abrasives. After a post cleaning, the particle contamination thus could be reduced due to the elec-trostatic repulsion. Our study proved that adding GCA into the U-H2O2slurries with BTA could further improve the Cu-CMP performance.

Keywords: Glycolic acid; Urea-hydrogen peroxide; Slurry; Copper; Chemical mechanical polishing

1. Introduction

Chemical mechanical polishing (CMP) of metal is widely recognized as the key technique to obtain global planarization for copper multilevel

inter-connection in IC manufacturing [1]. However,

the planarization mechanisms of metal CMP in-volve complicated chemical reactions and physical actions. So it is difficult to control the CMP per-formance, especially for a high corrosion-sensitiv-ity metal like Cu. According to different slurry components, the planarization mechanisms of me-tal CMP can be summarized as the following two types. The first one is that the metal surface may be polished with a dissolution-type slurry in which no surface film forms. In this case, the CMP

*

Corresponding author. Tel.: +886228927154x8041; fax: +886228960255.

E-mail address:[email protected](T.-H. Tsai).

(2)

process proceeds by the mechanical removal and chemical dissolution of metal itself[2]. In the sec-ond type of mechanisms, a thin abradable layer is continuously formed on metal surface by the reactions of metal and slurry components. These slurry components may include oxidants, inhibi-tors and other ﬁlm-forming agents. The thin abradable layer is then scanned away rapidly by the mechanical action or dissolves in the slurry with a chelating agent. Once the mechanical pol-ishing process has stopped, a thin passive ﬁlm re-mains on the surface and inhibits the proceeding

of wet etching [3]. Therefore, controlling CMP

process to reach global planarization is much eas-ier due to this mechanism.

The chemical components and properties of slurry determine the Cu-CMP performance princi-pally. As a result, many studies devoted to the development of copper slurries with high planari-zation ability and CMP eﬃciency. According to the second mechanism mentioned above, oxidants with high oxidizing ability and low contamination should be required to form a passive ﬁlm on Cu

surface. Hydrogen peroxide (H2O2) is a widely

used example at present [4,5]. However, H2O2

decomposes to H2O and O2easily, and this

decom-position will result in the slurry transport issues, unstable CMP performance and short shelf

life-time. Recently, urea-hydrogen peroxide

(U-H2O2) has been proposed as the oxidant for

Cu-CMP slurries due to its less self-decomposition

and better oxidizing performance during CMP[6].

Also, Kaufman and Kistler[7]reported that

or-ganic acids can be added into metal slurries to en-hance CMP performance by acting as pH adjusters or chelating agents. On the other hand, many researchers have examined the further eﬃciency of organic acids, and it was found that the action of organic acids would depend on the diﬀerent CMP conditions. For instance, in 1999, Hu et al.

[8]indicated that citric acid can inhibit Cu

corro-sion in HNO3slurries, and Zhang et al.[9]found

that citric acid can prevent the alumina-particle deposition on tungsten surface. Even though many organic acids, such as acetic acid [10], oxalic acid

[11], ethylenediamine tetraacetic acid (EDTA)

[12] and amino acid[13,14]have been extensively

studied, glycolic acid (GCA) has never been used.

As known, GCA is easy handling, water-soluble, biodegradable, non-volatile and can eﬃciently che-late the metal. Therefore, the objective of this pa-per is to investigate the eﬀect of GCA added into

U-H2O2 slurries on electrochemical behaviors of

Cu CMP. In addition to the measurement of polarization curves, the electrochemical impedance spectroscopy (EIS) was also obtained, and the equivalent circuit of surface reactions during Cu CMP was proposed. Although the equivalent cir-cuits of Al have been discussed[15], fewer studies examined that in Cu CMP. In this study, the spe-ciﬁc role of GCA and the information of the

equiv-alent circuit for CMP in U-H2O2-based slurries

were explored. Our experiments proved that GCA could improve Cu-CMP eﬃciency, and the planarization mechanisms of Cu CMP could be described well by the electrochemical methods.

2. Experimental

Commercial pure Cu sheets with 0.3-mm thick-ness were used, and the dimension of each

speci-men was 1· 1 cm2 for electrochemical

measuring. In addition, 3-in. wafers with an elec-troplated Cu ﬁlm were polished to obtain removal rates and surface roughness. All specimens were degreased by the cathode electrochemical method at 6 V for 20 s, and then were cleaned in 3 wt%

H2SO4solution followed by deionized (DI) water

rinsing. Afterward, the samples were dried in nitrogen gas and transferred to the electrochemical measuring equipment or polishing machine. The slurry was prepared by use of 50-nm alumina par-ticles (a-Al2O3), and its solid concentration in the

slurry was fixed at 5 wt% for all of the experi-ments. To investigate the effects of different chem-ical components on Cu CMP, we prepared the slurries by analytical grade reagents, including

urea-hydrogen peroxide (U-H2O2), benzotriazole

(BTA) and glycolic acid (GCA).

For the purpose to compare the stability of solutions, 5 wt% U-H2O2and 1.8 wt% H2O2, with

the equal level of peroxide in solutions, were peri-odically analyzed by titration with 1 wt% potas-sium permanganate solution to determine the peroxide activity. An acoustic spectrometer, model

(3)

DT-1200, was used to determine the zeta poten-tials of materials in diﬀerent pH values adjusted

by HNO3 or NH4OH. Dipping experiments were

conducted to determine the particle contamination on Cu surface. For dipping experiments, 300-nm

a-Al2O3 particles were used in slurries for SEM

observation. After dipped in slurries for 30 s, Cu samples were rinsed gently with DI water prior to SEM study.

The electrochemical test-cell was composed of Cu working electrode, platinum counter electrode and Ag/AgCl reference electrode, with a Luggin probe. The downward pressure and the rotating speeds during abrasion were 5 psi and 100 rpm, respectively. The dc electrochemical experiments were conducted by use of a potentiostat of EG&G, model 273A, while the corrosion software of EG&G, model 352, was adopted for electro-chemical calculations. The dc polarization curves with a voltage scan rate of 5 mV/s were selected to measure the corrosion current densities and po-tential. Meanwhile, the corrosion current density was converted to the corresponding corrosion rate (CR) in nm/min according to FaradayÕs law. In addition, EIS measuring was performed with an electrochemical interface system of Solartron, FRA 1260 and 1286, and a small amplitude per-turbation of 10 mV in a sinusoidal wave was

ap-plied with the operated frequency, 105–102 Hz,

to ensure the linearity for impedance analysis. For polishing experiments, a Lapmaster LM-51 polisher, operated at 80 rpm/60 rpm of the platen/

carrier rotating speeds, was used with the polishing pad of IC 1000/SUBA IV supplied by Rodel. Be-fore CMP, the slurry had to be stirred in a mixer to maintain the slurry suspension for the test per-iod. During CMP, the slurry was delivered to the gap between Cu and the pad with a ﬂow rate of 100 ml/min. The removal rate (RR) was calculated from the weight loss of Cu sheets before and after 5-min polishing, and the results were obtained by the average of three tests. Before or after CMP, the surface morphology of specimens was charac-terized by an atomic force microscope (AFM), and the surface rms-roughness (Rq) was calculated

by the software package.

3. Results and discussion 3.1. Durability of slurries

As mentioned above, the oxidant acts an impor-tant role of ﬁlm-formation in CMP slurry. In this

study, we chose two peroxides, H2O2 and

U-H2O2, as the oxidants. Table 1 lists the active

percentages of H2O2and U-H2O2, relative to their

initial values, and the pH values in various hydro-gen peroxide-based solutions. The results indicated that the active peroxide percentage in 1.8 wt% H2O2solution degraded at a faster rate than that

in 5 wt% U-H2O2 solution, but their pH values

were similar. Additionally, the degradation rates of the active peroxide percentage showed large

Table 1

Changes of active peroxide percentage (%) and pH value with time in various H2O2or U-H2O2slurries

Time (h) 0 10 50 100 250 500 1.8 wt% H2O2 % 100.0 96.35 75.16 57.42 41.99 33.44 pH 5.56 5.52 5.31 5.45 5.03 5.18 1.8 wt% H2O2+ 1 wt% NH4OH % 100.0 23.75 20.12 18.15 16.41 15.08 pH 10.03 10.21 9.78 9.65 9.41 9.32 1.8 wt% H2O2+ 1 wt% GCA % 100.0 96.01 83.24 71.31 64.02 57.88 pH 1.98 2.03 2.07 2.13 2.02 2.21 5 wt% U-H2O2 % 100.0 98.52 93.02 91.04 89.75 87.48 pH 5.87 5.65 5.89 5.76 5.81 5.92 5 wt% U-H2O2+ 1 wt% NH4OH % 100.0 76.12 52.43 48.21 35.02 31.80 pH 10.62 10.58 10.64 10.47 10.32 10.20 5 wt% U-H2O2+ 1 wt% GCA % 100.0 97.82 96.01 94.23 93.31 91.59 pH 2.04 2.10 2.08 2.11 1.94 2.07

(4)

variations when the U-H2O2 or H2O2-based

solutions were mixed with NH4OH or GCA. In

Cu-CMP slurry, NH4OH has been a well-known

chelating agent. However, it was found the active

peroxide percentage of 5 wt% H2O2solution with

1 wt% NH4OH reduced to 23.75% of its initial

value after only 10 h. Even though H2O2was

re-placed by U-H2O2, the peroxide activity of 5

wt% U-H2O2solution with 1 wt% NH4OH also

re-duced to 31.8% after 500 h. On the contrary, the addition of GCA, another chelating agent for Cu, could enhance the durability of slurries. After 500 h, the active peroxide percentage still exhibited a high level, 91.59%, for the 5 wt% U-H2O2

solu-tion with 1 wt% GCA. Also, the 5 wt% H2O2

solu-tion with 1 wt% GCA had higher peroxide activity

than that with NH4OH. This showed that using

GCA as a chelating agent instead of NH4OH

could form a stable solution with a longer shelf lifetime for hydrogen peroxide-based CMP slurry. 3.2. Polarization curves

The abrasion eﬀect on the polarization curves

of Cu in various slurries are shown in Fig. 1and

Fig. 2, and their corresponding corrosion parame-ters and Cu removal rates (RR) are all listed in

Table 2. The ﬁrst thing noticeable is the diﬀerence

between polarization curves of Fig. 1, especially

the corrosion potential drop after abrasion

(DEd). It is obvious that DEdin 1 wt% GCA slurry

was smaller than that in 5 wt% U-H2O2. As

re-ported [16], the potential drop is attributed to

the change on sample surface from the formation of passive ﬁlm to its removal. Therefore, as shown in Fig. 1, the larger DEd in 5 wt% U-H2O2

illus-trated a faster formation and a quicker removal of the passive ﬁlm on Cu surface. On the other hand, Cu in 1 wt% GCA with a higher anodic cur-rent density, relative to 5 wt% U-H2O2, exhibited a

faster dissolution rate of Cu due to the chelating action between Cu surface and GCA. In further comparison with the corrosion parameters in

Table 2, the corrosion potential and corrosion

cur-rent density were both higher in 5 wt% U-H2O2

than those in 1 wt% GCA. This was attributed to the Cu with higher oxidation state and greater oxidizing rate in U-H2O2 solution. Thus, it could

be shown again that U-H2O2owned a strong

oxi-dizing action for Cu, and the formed passive ﬁlm could protect Cu from deeper corrosion.

When 5 wt% U-H2O2 was mixed with 1 wt%

GCA, DEd decreased drastically but the anodic

current density of Cu increased as shown in

10-7 1x10-5 10-3 -400 -200 0 200 400 600 1 w% GCA 5 w% U-H₂O₂ no abrasion abrasion E (mV vs Ag/AgCl ) i (A/cm2)

(5)

Fig. 2. It resulted from the thinning passive ﬁlm on Cu surface by the chelating action of GCA, so the Cu consumption was enhanced by both chelating and polishing. Thus, as listed inTable 2, the high corrosion rate, 19.59 nm/min, and removal rate, 605.82 nm/min, were obtained in the mixed slurry,

5 wt% U-H2O2 and 1 wt% GCA. Furthermore,

Fig. 2also shows that DEdincreased substantially

by adding 0.1 wt% BTA into 5 wt% U-H2O2and 1

wt% GCA slurry even though the current density still maintained a low level. It indicates that Cu

in 5 wt% U-H2O2 slurry with 1 wt% GCA and

0.1 wt% BTA showed the high mechanical sensitiv-ity and good anti-corrosion capabilsensitiv-ity. As a result, the Cu removal rate attained 594.14 nm/min, but

the corrosion rate reduced to 3.37 nm/min in this mixed slurry. The AFM micrographs of Cu sur-face before and after CMP with this slurry are

shown in Fig. 3(a) and (b), respectively. In

com-parison with the AFM images, it shows the surface roughness could be reduced eﬃciently after CMP

with the proposed slurry, 5 wt% U-H2O2, 1 wt%

GCA and 0.1 wt% BTA. 3.3. EIS results

The EIS results of Cu in the slurry of 5 wt%

U-H2O2, 1 wt% GCA and 0.1 wt% BTA are

pre-sented by Nyquist and Bode plots as shown in

Fig. 4(a) and (b), respectively. When Cu was

10-7 1x10-5 10-3 0 300 600 with 0.1 w% BTA 5 w% U-H 2O2 + 1 w% GCA no abrasion abrasion E (mV vs Ag/AgCl ) i (A/cm )2

Fig. 2. Eﬀects of abrasion and BTA on polarization curves for Cu in 5 wt% U-H2O2with 1 wt% GCA slurries (abrasion: 5 psi and 100 rpm).

Table 2

Summary of corrosion parameters and removal rates (RR) of Cu in various slurries

Slurry No abrasion Abrasion DEda(mV) RR (nm/min) CR (nm/min)

Ecorr(mV) icorr(A/cm2) Ecorr(mV) icorr(A/cm2)

5 wt% U-H2O2 153 1.13· 104 46 1.33· 104 107 472.63 2.92 1 wt% GCA 8 1.42· 105 ₃ _4.48_{· 10}5 ₁₁ _184.30 _1.07 5 wt% U-H2O2+ 1 wt% GCA 182 4.60· 104 145 8.85· 104 37 605.82 19.59 5 wt% U-H2O2+ 1 wt% GCA + 0.1 wt% BTA 323 9.63· 105 ₁₉₁ _1.53_{· 10}4 ₁₃₂ _594.14 _3.37 a _DE

(6)

immersed in the mixed slurry, a dual non-ideal semicircle response of Nyquist plot, shown in

Fig. 4(a), could be found. As Griﬃn et al. [17]

pointed out, this response is caused from the for-mation of the secondary conduction paths or dis-continuities within the copper oxide layer at or near the incoherent precipitates. Thus, the con-stant-phase-element (CPE) was adopted in the equivalent circuit to ﬁt the behaviors at the inter-face of Cu and slurry[6]. Furthermore, it has been reported the inductor response in the low fre-quency range is due to the adsorption or the

evo-lution of gas on surface [18]. Therefore, our

equivalent circuit for describing Cu CMP would also contain the inductor element. In summary of the above facts, the electrochemical reactions at the interface of Cu and slurry would be simulated by the equivalent circuit ofFig. 5. As shown in this ﬁgure, the equivalent circuit consists of solution

resistance (Rs), double-layer CPE (CPEdl),

charge-transfer resistance (Rct), passive-ﬁlm CPE

(CPEpf), passive-ﬁlm resistance (Rpf) and

adsorp-tion inductance (Lad). The ﬁtting results by use

of CPEs, presented by the solid lines inFig. 4(a) and (b), would approach those experiment results rather than those by use of capacitors.

The results are also displayed by the Bode plot in Fig. 4(b). According to ac impedance theory

[18], the impedance at frequency A and B,

indi-cated by the dashed lines of Fig. 4(b), can be

re-garded as Rs+ Rpf+ Rct, and Rs+ Rpf,

respectively, and the impedance at suﬃciently high frequency, 105Hz, is Rs. Thus, we could calculate

these resistances from the Bode plot. It was found

that Rsapproached 78.88 X cm2when Cu was

im-mersed into the mixed slurry, 5 wt% U-H2O2, 1

wt% GCA and 0.1 wt% BTA. The polishing, how-ever, increased Rsto 87.28 X cm2due to the

smal-ler ﬂuid pathway between Cu surface and the pad. In addition, rise in Rpfand Rctcould be found with

the increasing immersion time. The result indi-cated that the adsorbed BTA-layer and the formed oxide ﬁlm inhibited the reactions continuously, and this caused the resistance to increase. As a

Fig. 3. AFM micrographs of Cu surface with roughness (Rq): (a) before CMP, and (b) after CMP with the mixed slurry, 5 wt% U-H2O2, 1 wt% GCA and 0.1 wt% BTA.

(7)

mechanical action of 5 psi and 100 rpm was ap-plied, the adsorbed layers and passive ﬁlms were

destroyed, and then Rpf and Rct reduced

substantially.

Moreover, the phase angle in Fig. 4(b) also

shows that the high-frequency peak was related to the passivation while the low-frequency peak was aﬀected by the surface reactivity. The former was lowered in most of the CMP duration, and the latter was raised with the increasing immersion time. Also, the positive phase-angle value, depend-ing on the adsorption, increased with the immer-sion time and disappeared during CMP. All of

the above results indicated that the formation of gas, the adsorption of BTA-layer and the genera-tion of oxide ﬁlm on the Cu surface could be illus-trated well by the proposed equivalent circuit, and also the behaviors sensitive to the mechanical force could be investigated.

3.4. Relationship between CPE and surface roughness

In our observations, the passive ﬁlm on Cu sur-face was a loose oxide layer, and its sursur-face was of-ten rough and porous. So a non-ideal capacitor

10-1 100 101 102 103 104 105 5 0 -5 -10 -15 -20 B A 5 w% U-H 2O2 + 1 w% GCA + 0.1 w% BTA 1 min immersion 3 min immersion 6 min immersion during CMP

Phase Angle (degree)

Frequency (Hz) 80 100 200 400 600 800 1000 Impedance ( Ω -cm ) 2 100 200 300 400 0 -100 -200 -300 increasing ω Z''( Ω -cm 2) Z'(Ω-cm )2

5 w% U-H₂O₂ + 1 w% GCA + 0.1 w% BTA

1 min immersion 3 min immersion 6 min immersion during CMP (a) (b)

Fig. 4. Eﬀects of immersion time and CMP on: (a) Nyquist plot and (b) Bode plot for Cu in 5 wt% U-H2O2, 1 wt% GCA and 0.1 wt% BTA slurry. Simulated results are represented by the solid curves (CMP: 5 psi and 100 rpm).

(8)

was induced into the equivalent circuit for simulat-ing the interface behaviors of Cu and slurry. The CPE herein was substituted for the ideal capacitor.

The CPE impedance is written as[19],

ZCPE ¼ ½T ðjxÞn1;

where T is a general admittance function, j is the complex operator ðpffiffiffiffiffiffiffi1Þ, and n is an adjustable parameter that usually lies between 0.5 and 1. When n = 1, the CPE describes an ideal capaci-tance with the high planarization or

homogene-ities. For the case of n = 0.5, a Warburg

impedance relation is obtained; this impedance is associated with the concentration and the diﬀu-sion-related processes. Generally, the deviation from the ideal surface ﬁlm, i.e. ideal capacitor, can be estimated by this adjustable parameter.

Fig. 6, containing the Cu-CMP results by use of U-H2O2-based slurries in Ref. [6], illustrates the

dependence between the adjustable parameters and the roughness. In which, the parameters such as double-layer (ndl) and passive ﬁlm (npf) were

in-cluded, and the roughness (Rq) was calculated by

the AFM technique. Obviously, it exhibited that all of the npfvalues were lower than the ndlvalues

and independent of the roughness. On the con-trary, ndlseemed further relevant to the Cu surface

roughness (Rq). The reason was evident that CPEpf

was involved in the porous passive layer while CPEdlwas aﬀected only by the Cu surface. As a

re-sult, the ndl value could provide a good index to

monitor the planarization-performance on Cu surface. It was also found inFig. 6that the lowest surface roughness, 2.64 nm, was obtained in the

U-H2O2slurry mixed with 0.1 wt% BTA and 1 wt%

NH4OH instead of GCA. Even though the slurry

containing GCA was not the best choice for pla-narity, the addition of GCA still could enhance the slurry stability. As reported[20], a stable slurry is important for good CMP performance because it can not only improve CMP reliability but also reduce the defects of the surface after CMP. 3.5. Surface potential and particle contamination

Fig. 7 shows the zeta potentials of Cu and

a-Al2O3 particles as a function of pH in 0.1 wt%

BTA solutions with or without 1 wt% GCA. The corresponding isoelectric points (IEPs), the pH of

zero charge, of Cu and a-Al2O3without GCA were

0.725 and 9.109, respectively. This meant that the Cu and a-Al2O3surfaces had opposite charges for

pH values from 0.725 to 9.109. The opposite charges resulted in the attraction, and then

a-Al2O3 particles would be easily deposited on Cu

surface and diﬃcultly removed. However, when 1 wt% GCA was added to the solutions, the IEPs of Cu and a-Al2O3reduced to 3.672 and 6.253,

respec-tively. Obviously, the pH range of Cu and a-Al2O3

with opposite charges was shortened as shown in

Fig. 7. This indicates that a-Al2O3deposition could

be inhibited by use of GCA due to the electrostatic

(9)

0 2 4 6 8 10 12 -60 -40 -20 0 20 40 60 Zeta Potential (mV) pH α -Al2O3 Cu 1 w% GCA α -Al2O3 Cu

Fig. 7. Zeta potentials of a-Al2O3and Cu as a function of pH in 0.1 wt% BTA solutions with or without 1 wt% GCA.

0.6 0.8 1.0 0 2 4 6 8 Roughness (nm) n n_pf n_dl 5 w% U-H₂O₂ 5 w% U-H₂O₂ + 0.1 w% BTA 5 w% U-H₂O₂ + 0.1 w% BTA + 1 w% NH₄OH 5 w% U-H₂O₂ + 0.1 w% BTA + 1 w% GCA

Fig. 6. Surface roughness vs. the adjusted parameters of passive ﬁlm, npf(ﬁlled symbol) and double layer capacitances, ndl(hollow symbol) in various slurries for Cu CMP at 5 psi and 100 rpm.

(10)

repulsion. Fig. 8 displays the SEM images of a-Al2O3contamination, shown by those white points,

on Cu surface. It shows that a clearer Cu surface, with less a-Al2O3deposition, was obtained by use

of the slurry with 1 wt% GCA. These SEM images supported the above results of zeta potential.

4. Conclusions

In this study, the eﬀect of glycolic acid (GCA) on Cu CMP was investigated by electrochemical methods, including titration, polarization and impedance experiments. The results showed that GCA could increase the stability of slurries by

pre-venting the hydrogen-peroxide component from rapid decomposition. The electrochemical studies from polarization and impedance measurements veriﬁed that GCA could chelate with Cu and cop-per oxide on the polished surface in the existence of oxidants and inhibitors. Thus, the Cu-CMP per-formance could be enhanced by use of GCA, and the Cu removal rate could achieve 594.14 nm/min with a low surface roughness, 6.09 nm, in the mixed

slurry, 5 wt% U-H2O2, 1 wt% GCA and 0.1 wt%

BTA. In addition, a deliberate equivalent circuit for Cu-CMP system was proposed. The ﬁtting re-sults of EIS showed that the adjustable parameter of double-layer CPE (ndl) would depend on the

sur-face roughness and could be regarded as an index

Fig. 8. SEM images of Cu samples dipped for 30 s in 5 wt% a-Al2O3(300 nm) and 0.1 wt% BTA slurries: (a) without GCA; (b) with 1 wt% GCA.

(11)

to CMP planarization. The slurry with GCA still could result in the electrostatic repulsion between

Cu surface and a-Al2O3 over a pH range larger

than that without GCA. Therefore, the addition of GCA into copper slurries could eﬃciently de-crease particle contamination and easily enhance the post-CMP cleaning process.

Acknowledgements

The authors would like to thank the National Science Council in Taiwan for ﬁnancially support-ing this research under Contract No. NSC 93-2214-E-149-001. Also, the authors are pleased to acknowledge the help extended by Northern Taiwan Institute of Science and Technology and National Taiwan University.

References

[1] H. Xiao, Introduction to Semiconductor Manufacturing Technology, Prentice-Hall, Inc., New Jersey, 2001. [2] J.M. Steigerwald, R. Zirpoli, S.P. Murarka, D. Price, R.J.

Gutmann, J. Electrochem. Soc. 141 (1994) 2842. [3] F.B. Kaufman, D.B. Thompson, R.E. Broadie, M.A. Jaso,

W.L. Guthrie, D.J. Pearson, M.B. Small, J. Electrochem. Soc. 138 (1991) 3460.

[4] M. Ziomek-Moroz, A. Miller, J. Hawk, K. Cadien, D.Y. Li, Wear 255 (2003) 869.

[5] T. Du, Y. Luo, V. Desai, Microelectron. Eng. 71 (2004) 90.

[6] T.H. Tsai, Y.F. Wu, S.C. Yen, Appl. Surf. Sci. 214 (2003) 120.

[7] V.B. Kaufman, R.C. Kistler, US Pat. 5,954,997, 1999. [8] T.C. Hu, S.Y. Chiu, B.T. Dai, M.S. Tsai, I.C. Tung, M.S.

Feng, Mater. Chem. Phys. 61 (1999) 169.

[9] L. Zhang, S. Raghavan, S. Meikle, G. Hudson, J. Electrochem. Soc. 146 (1999) 1442.

[10] J.M. Steigerwald, S.P. Murarka, R.J. Gutmann, D.J. Duquette, Mater. Chem. Phys. 41 (1995) 217.

[11] K. Yang, S. Avanzino, C.M.C. Woo, US Pat. 6,143,656, 2000.

[12] S. Aksu, F.M. Doyle, J. Electrochem. Soc. 149 (2002) B340.

[13] M. Hariharaputhiran, J. Zhang, S. Ramarajan, J.J. Kele-her, Yuzhuo Li, S.V. Babu, J. Electrochem. Soc. 147 (2000) 3820.

[14] S. Aksu, F.M. Doyle, J. Electrochem. Soc. 149 (2002) B352.

[15] H.S. Kuo, W.T. Tsai, J. Electrochem. Soc. 147 (2000) 149.

[16] T.H. Tsai, S.C. Yen, Appl. Surf. Sci. 201 (2003) 190. [17] A.J. Griﬃn, F.R. Brotzen, C.F. Dunn, J. Electrochem.

Soc. 141 (1994) 3473.

[18] J.R. Macdonald, Impedance Spectroscopy, Wiley, New York, 1987.

[19] R. Babic, M.M. Hukovic, A. Jukic, J. Electrochem. Soc. 148 (2001) B146.

[20] L. Zhang, S. Raghavan, M. Weling, J. Vacuum, Soc. Technol. B 17 (1999) 2248.