Mechanism of copper deposition in a sulphate bath containing chlorides

The mechanism of copper deposition in a sulphuric acid–sulfate bath in the ... the chloride ions added to a sulphuric acid–sulfate copper-plating bath have a.
358KB taille 25 téléchargements 322 vues
Journal of

Electroanalytical Chemistry Journal of Electroanalytical Chemistry 572 (2004) 367–375 www.elsevier.com/locate/jelechem

Mechanism of copper deposition in a sulphate bath containing chlorides C. Gabrielli *, P. Mocßoteguy, H. Perrot, R. Wiart UPR15 du CNRS, LISE, Universite Pierre et Marie Curie, 4 Place Jussieu, Paris cedex 5, 75252, France Received 14 July 2003; received in revised form 12 January 2004; accepted 12 January 2004 Available online 6 May 2004

Abstract The mechanism of copper deposition in a sulphuric acid–sulfate bath in the presence of chlorides was investigated by electrochemical impedance measurements. A model where the chloride ions added to a sulphuric acid–sulfate copper-plating bath have a catalytic action not limited by mass transport has been proposed. This shows that the observed mass transport limitation is due to the diffusion of cupric ions and that, at high current densities, the copper deposition is mainly due to a mechanism involving CuCl formation on the electrode. Ó 2004 Elsevier B.V. All rights reserved. Keywords: Copper; Electrocrystallization; Chloride; Electrochemistry; Impedance

1. Introduction From galvanoplasty to printed circuits, acid galvanic copper deposition baths have been used in industry for generations. Now, copper is replacing aluminum in microelectronic interconnects, leading to a new field of application of this technique. The demanding performances of copper deposition such as in the damascene process needs a thorough knowledge of the reaction mechanism underlying this phenomenon. Cathodic deposition of copper in sulphuric acid solution is generally supposed to occur through two consecutive charge transfer steps involving the soluble intermediate Cuþ [1–4]. Impedance techniques have been used to investigate copper deposition; they have revealed, in addition to the expected charge transfer resistance, low frequency relaxations [5,6], which have been ascribed to various processes. The low frequency capacitive features have been interpreted as diffusion of Cu2þ from the solution and reaction intermediate relaxations. The inductive *

Corresponding author. Tel.: 33-1-44-27-41-53; fax: 33-1-44-27-40-

74. E-mail address: [email protected] (C. Gabrielli). 0022-0728/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jelechem.2004.01.025

features have been interpreted as the relaxation of the surface concentration of adatoms [7] and as the activation of the electrode area with increasing potentials although it has not been possible to make a clear distinction between a slow increase of the surface area due to the nucleation and the development of growth centers [8] and a slow removal of inhibiting species such as anionic species, hydroxides and organic molecules [6]. A previous thorough investigation of copper deposition [9] has already shown that electrode activation occurs with increasing current density and that the low frequency features appeared to be strongly dependent on the growth mode of deposit. The lowest frequency inductive loop of the impedance has been interpreted in terms of relaxation of the electrode area on the basis of the birth and growth of monolayers formed on the facets of the crystallites. It has been shown that the addition of low quantities of chlorides modifies the electrode kinetics. The interest in copper deposition has been renewed recently owing to its use in the damascene process for copper interconnects in microelectronics [10,11]. This process needs the addition of several additives in the plating bath; however to understand the deposition mechanism in this situation a thorough investigation of the action of chlorides has been undertaken. Investiga-

368

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

tions by voltammetry combined with radiotracer adsorption [12] or electrogravimetry by using a quartz crystal microbalance [13] concluded to the formation of a CuCl layer on the copper electrode surface. The latter is in favour of a catalytic action of the chloride in the formation of CuCl on the copper surface at a rate controlled by the diffusion of chloride from the electrolyte. This catalytic action of Cl has been also demonstrated in the copper deposition in 0.5 M HClO4 solutions [14]. This is also supported by the fact that despite the possibility of a CuCl film, SIMS results confirm that no substantial inclusion of chloride and sulphate is found in the copper deposit [15]. However, it has been noticed that the apparent Cl effect can be reversed from inhibition to activation as the electrolyte acidity increases [16]. Usually, the deposition bath contains Cu2þ ions in high concentrations and low concentrations of chlorides. They are supposed to act as catalysts. The aim of this paper is to investigate the role of chlorides in the copper deposition mechanism by kinetic methods based on impedance measurement techniques.

Then the equations, which govern the copper deposition, are:

2. Theory

ki ¼ ki0 ebi E

Cupric ions are supposed to be reduced by the following two different paths. The first is a direct discharge of the ions in two steps: k1

Cu2þ þ e ¢ Cuþ

ð1Þ

k1

k2

Cuþ þ e ¢ Cu

ð2Þ

k2

The second follows a parallel path through a catalytic complexation with chloride ions, and adsorption of copper chloride on the copper surface: k3

Cu2þ þ Cl þ e ¢ CuClads k3

k4

CuClads þ e ¢ Cu þ Cl k4

ð3Þ

k5

k5

ð6Þ b1

b2

dh1 ¼ k3 ð1  h1  h2 Þc0  k3 h1  k4 h1 þ k4 dt  ð1  h1  h2 Þ; dh2 ¼ k5 ð1  h1  h2 Þ  k5 h2 ; dt

ð7Þ ð8Þ

IF ¼  F ½k1 ð1  h1  h2 Þc0  k1 ½Cuþ  þ k2 ½Cuþ ð1  h1  h2 Þ  k2 ð1  h1  h2 Þ þ k3 ð1  h1  h2 Þc0  k3 h1 þ k4 h1  k4 ð1  h1  h2 Þ þ k5 h2  k5 ð1  h1  h2 Þ;

ð9Þ

where h1 , h2 are the surface coverages of CuClads and Aads , respectively, [Cuþ ] is the concentration of Cuþ on the electrode, and cðxÞ is the concentration of Cu2þ at a distance x from the electrode. In the following cðx ¼ 0Þ will be called c0 . The rate constants are of the form: ð10Þ

with bi < 0 for cathodic reactions ðI < 0Þ and bi > 0 for anodic reactions (I > 0). As chloride ions are not consumed at the electrode, their concentration is considered as a constant and is involved in the rate constant k3 . If the direct reactions through k1 , k2 are supposed to be fast, the relaxation of [Cuþ ] with time is fast in comparison with those of the other parameters h1 , h2 and c0 , so that [Cuþ ] can be considered to vary in phase with the electrode potential E. Then at any time, one has k1 ð1  h1  h2 Þc0  k1 ½Cuþ   k2 ð1  h1  h2 Þ½Cuþ  þ k2 ð1  h1  h2 Þ ¼ 0; ð11Þ then ½Cuþ  ¼

ð4Þ

In addition, the slow adsorption/desorption of a blocking species, e.g., for an anion A , has to be considered to interpret the experimental data completely Aads þ e ¢ A :

d½Cuþ  ¼ k1 ð1  h1  h2 Þc0  k1 ½Cuþ  dt  k2 ð1  h1  h2 Þ½Cuþ  þ k2 ð1  h1  h2 Þ;

ðk1 c0 þ k2 Þð1  h1  h2 Þ k1 þ k2 ð1  h1  h2 Þ

ð12Þ

and IF ¼ F ½ð2k1 þ k3 Þð1  h1  h2 Þc0  2k1 ½Cuþ   k3 h1 þ k4 h1  k4 ð1  h1  h2 Þ þ k5 h2

ð5Þ

The Cuþ ion is considered as a soluble species whose electrode coverage has not to be considered. In addition, for simplicity, the two-step rates are assumed to be fast with regard to the diffusion flux of Cuþ ion away from the electrode, which can be disregarded in the mass balance of Cuþ .

 k5 ð1  h1  h2 Þ:

ð13Þ

The diffusion of the Cu2þ ions follows the classical diffusion law ocð xÞ o2 cð xÞ ¼D ot ox2 with the following boundary conditions:

ð14Þ

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

D

At the electrode surface (x ¼ 0)

and

ocð xÞ ¼ ðk1 þ k3 Þð1  h1  h2 Þc0  k1 ½Cuþ   k3 h1 ox ð15Þ

DIF ¼ R1 t DE þ F f½ð2k1 þ k3 Þc0  k4 þ k3  k4

and beyond the diffusion layer of thickness dðx P d), cðxÞ ¼ a constant. Now, the electrochemical processes are analyzed first at steady-state and second in a low amplitude dynamic regime. 2.1. Steady state (dh=dt ¼ 0)

 k5 Dh1  ½ð2k1 þ k3 Þc0  k4  k5  k5 Dh2 þ ð2k1 þ k3 Þð1  h1  h2 ÞDc0 g; R1 t ¼  F ½ðð2b1 k1 þ b3 k3 Þc0  b4 k4 Þð1  h1  h2 Þ  2b1 k1 ½Cuþ  þ ðb4 k4  b3 k3 Þh1 þ b5 k5 h2  b5 k5 ð1  h1  h2 Þ: The diffusion of Cu

þ k4 ð1  h1  h2 Þ ¼ 0;

ð16Þ

k5 ð1  h1  h2 Þ  k5 h2 ¼ 0;

ð17Þ

IF ¼ F ½ðð2k1 þ k3 Þc0  k4 Þð1  h1  h2 Þ  2k1 ½Cuþ  þ ðk4  k3 Þh1 ;

jxDcð xÞ ¼ D

ð18Þ

h1 ¼

follows

o2 Dcð xÞ ox2

k3 c0 þ k4 k3 c0 þ k4 þ ðk5 þk5kÞ5ðk3 þk4 Þ

ð30Þ

þ k3 Dh1 þ ðk1 þ k3 Þð1  h1  h2 ÞDcð0Þ;

ð19Þ

ð29Þ

with boundary conditions at x ¼ 0   oDc ¼ A3 DE  ðk1 þ k3 ÞðDh1 þ Dh2 Þc0 D ox x¼0

hence 1  h1 h2 ¼ k5 ; k5 þ k5

ð28Þ

where



k3 ð1  h1  h2 Þc0  k3 h1  k4 h1

369

ð31Þ

where A3 ¼ ðb1 k1 þ b3 k3 Þð1  h1  h2 Þc0  b1 k1 ½Cuþ 

;

and IF ¼ IF1 þ IF2 ;

ð21Þ

where the two partial currents, IF1 and IF2 , correspond to the first and second paths, respectively: IF1 ¼ F ½2k1 ð1  h1  h2 Þc0  2k1 ½Cuþ ;

 b3 k3 h1

ð20Þ

ð22Þ

IF2 ¼ F ½ðk3 c0  k4 Þð1  h1  h2 Þ þ ðk4  k3 Þh1 : ð23Þ 2.2. Electrochemical impedance The responses Dh1 , Dh2 , DIF and Dc ðx ¼ 0Þ to a small amplitude sine wave perturbation DE with pulsation x ¼ 2pf , are such as jxb1 Dh1 ¼ A1 DE  ðk3 c0 þ k3 þ k4 þ k4 ÞDh1  ðk3 c0 þ k4 ÞDh2 þ k3 ð1  h1  h2 ÞDc0 ; ð24Þ

Eq. (30) has the general solution: rffiffiffiffiffi pffiffiffi jx jx d D Dcð xÞ ¼ 2N  e ;  shð x  dÞ  D

From Eqs. (24) and (26)   A1  h2Að2xÞ ðk3 c0 þ k4 Þ DE þ k3 ð1  h1  h2 ÞDc0 ; Dh1 ¼ h1 ð x Þ ð35Þ Dh2 ¼

A2 DE  k5 Dh1 ; h2 ðxÞ

A1 ¼ b3 k3 ð1  h1  h2 Þc0  b3 k3 h1  b4 k4 h1

and

ð37Þ

ð25Þ

h2 ðxÞ ¼ jxb2 þ k5 þ k5 :

ð26Þ

Then, by substituting Dh1 and Dh2 by their values in Eq. (31), one has   oDc D ¼ A4 DE þ A5 Dc0 ; ð39Þ ox x¼0

where A2 ¼ b5 k5 ð1  h1  h2 Þ  b5 k5 h2

ð36Þ

where

where

jxb2 Dh2 ¼ A2 DE  k5 Dh1  ðk5 þ k5 ÞDh2 ;

ð33Þ

therefore, by using the boundary condition at x P d rffiffiffiffiffi rffiffiffiffiffi pffiffiffi oDcð xÞ jx d jxD jx ¼ 2N e : ð34Þ  chð x  dÞ  ox D D

h1 ðxÞ ¼ jxb1 þ k3 c0 þ k3 þ k4 þ k4 k5  ðk3 c0 þ k4 Þ jxb2 þ k5 þ k5

þ b4 k4 ð1  h1  h2 Þ;

ð32Þ

ð27Þ

ð38Þ

370

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

where

0

A4 ¼ A3  ð k 1 þ k 3 Þ @ 





A7 ¼ R1 t 

h1 ðxÞ 1



þ FA6

A2 A c0  k3 h2 ð x Þ   A1  h2Að2xÞ ðk3 c0 þ k4 Þ

 1

k5 h2 ðxÞ

A1  h2Að2xÞ ðk3 c0 þ k4 Þ

A1  A2 ðkh32cð0xþkÞ 4 Þ h1 ðxÞ

ð47Þ

;

  k3 A 8 ¼ F A6 þ ð2k1 þ k3 Þ ð1  h1  h2 Þ: h1 ð x Þ

þ

h1 ðxÞ

FA2 ½ð2k1 þ k3 Þc0  k4  k5  k5  h2 ð x Þ

ð40Þ

Finally DIF A4 A8 qffiffiffiffi ¼ A7  pffiffiffiffiffiffiffiffiffi ; DE jxD  coth d jx þ A

and

D



k3 ð1  h1  h2 Þ A5 ¼  ðk1 þ k3 Þ h1 ðxÞ   k5  1 c 0 þ ð k 1 þ k 3 Þ ð 1  h1  h2 Þ h2 ðxÞ ð 1  h1  h2 Þ ;  k3 k3 h1 ð x Þ

ZF !

ð48Þ

i.e.,

ð49Þ

5

pffiffiffi jx p 1 þ A5 ffiffiffiffiffiffiD thd

ZF ¼ ð41Þ

1 A7

jxD

  thdpffiffijxffi : 1 þ A5  AA4 A7 8 pffiffiffiffiffiffiD jxD

Notice that if diffusion is very fast (D ! 1)

1 h2 ð x Þ ¼ : 1 A7 Rt  h2 ðxÞ  FA2 ½ð2k1 þ k3 Þc0  k4  k5  k5  þ FA6 A1 h2 ðxÞA2 ðk3 c0 k4 Þ h1 ðxÞ

therefore from Eq. (34)

ð50Þ

ð51Þ

Two particular cases are of interest: (i) no A adsorption: k5 ¼ k5 ¼ 0, and diffusion very fast, then h2 ¼ 0,

A4 DE N ¼  pffiffiffiffiffiffiffiffiffi pffiffiffi  qffiffiffiffi qffiffiffiffi jx jx jx d D 2 jxD  chd: D þ A5  shd D  e

ZF ! ð42Þ

1 1 ¼ 1 A7 Rt þ FA6 h Að1xÞ 1

where R1 t ¼ F ½ð2b1 k1 þ b3 k3 Þð1  h1 Þc0 þ b4 k4 h1 :

and A4 DE qffiffiffiffiffiffiffiffiffiffiffiffiffiffi : Dc ðx ¼ 0Þ ¼  pffiffiffiffiffiffiffiffiffi þ A5 jxD  coth d jx D

ð43Þ

At last, by substituting Dh1 , Dh2 and Dc0 by their values in Eq. (28), one has:   FA2 ½ð2k1 þ k3 Þc0  k4  k5  k5  DIF ¼ R1  DE t h2 ð x Þ þ F ½A6 Dh1 þ ð2k1 þ k3 Þð1  h1  h2 ÞDc0 ; ð44Þ where A6 ¼ ½ð2k1 þ k3 Þc0  k4 þ k3  k4  k5  þ ½ð2k1 þ k3 Þc0  k4  k5  k5 

k5 h2 ðxÞ

ð45Þ

then DIF ¼ A7 DE þ A8 Dc0 ; where

ð46Þ

(ii) Copper deposition without chlorides and A adsorption: k3 ¼ k4 ¼ k5 ¼ k5 ¼ 0 qffiffiffiffi 1 0 thd: jx @1 þ k1 pffiffiffiffiffiffiffiffiDffi A: ZF ¼ R1 t jxD Simulations of the I–V curves (Fig. 1) and electrochemical impedances (Fig. 2) were carried out for the parameters given in the caption of Fig. 1. Fig. 1(A) shows the changes of the steady-state total current I and partial currents I1 and I2 with respect to the potential. This shows that at low overvoltages the first path, through Cuþ , is preponderant whereas at higher overvoltages the second path, through CuCl, is the major source of copper deposition. Fig. 1(B) shows the product Rt I with respect to the potential; it increases when the overvoltage increases, with a maximum value attained at a potential close to )0.35 V. The calculated impedances were plotted without (Fig. 2(A)) and with (Fig. 2(B)) mass transport limita-

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

371

1 0,08

0,01

0,06

1E-3

R tI / V

Current density / A cm

-2

0,1

1E-4 1E-5

I1 I2 I = I1 + I2

1E-6

0,04

0,02

0,00

1E-7 -0,6

-0,5

(A)

-0,4

-0,6

-0,3

-0,5

(B)

Potential / V

-0,4

-0,3

Potential / V

Fig. 1. Calculated steady-state quantities of copper deposition for the parameter values: k10 ¼ 1011 cm s1 , b1 ¼ 37 V1 , k20 ¼ 2  106 cm s1 , b2 ¼ 10 V1 , k30 ¼ 22  1013 cm s1 , b3 ¼ 43 V1 , k40 ¼ 1013 mol cm2 s1 , b4 ¼ 30 V1 , k50 ¼ 2:5  1012 cm s1 , b5 ¼ 15 V1 , k10 ¼ 5  1011 mol cm2 s1 , b1 ¼ 15 V1 , k20 ¼ 0, k30 ¼ 22  1010 mol cm2 s1 , b3 ¼ 2 V1 , k40 ¼ 0, k50 ¼ 0, b1 ¼ b2 ¼ 108 cm, D ¼ 5  107 cm2 s1 , Cd ¼ 105 lF cm2 at E ¼ 0:4 V. (A) Current–voltage curves where I ¼ I1 þ I2 in the log-linear plot (the absolute values of the current densities are plotted in this figure). (B) Change of the product Rt I with respect to the potential. 50

40

40

20

1000

10

100

10 1

0

0.001

0.1

0.01

-10

-Imaginary Part Z(ω) / Ω

-Imaginary Part Z(ω) / Ω

30

0.1

20 10

1 100

0

10 0.001

-10

0.01

-20 -30

-20 0

(A)

30

10

20

30

40

50

0

60

Real Part Z(ω) / Ω

(B)

10

20

30

40

50

60

70

80

Real Part Z(ω) / Ω

Fig. 2. Calculated electrochemical impedances (A) without mass transport limitation, (B) with mass transport limitation.

tion. The high frequency capacitive loop is related to the charge transfer resistance in parallel to the double layer capacity. The second capacitive loop in the middle frequency range is related to the CuCl adsorption. The third capacitive loop at lower frequencies (Fig. 2(B)) is related to the Cu2þ diffusion and does not exist when mass transport is not limiting, i.e., when the rotation velocity of the copper disc is sufficiently high. Finally, the low frequency inductive loop is related to the desorbing species.

3. Experimental The electrolytes were made up with analytical grade purity chemicals. The copper deposition was carried out in 1.8 M H2 SO4 and 0.25 M CuSO4 aqueous solutions

where chlorides were added and various concentrations of NaCl. The counter electrode was a large area copper plate (Goodfellow, impurity max. 10 ppm). The polycrystalline working electrode was a 5 mm diameter copper rotating disc. Before each experiment the working electrode was polished with emery paper up to 1200 grade. Finally, the specimens were washed in distilled water. The current–voltage curves were recorded potentiostatically (Autolab PGSTAT 100) using a 0.1 mV s1 voltage sweep in the negative direction from the rest potential to )0.6 V vs SSE. The complex electrode impedance was measured from 60 kHz to 103 Hz with 5 measured frequencies per decade and 20 mV peak to peak sine wave amplitude signals using a Solartron 1250 frequency response analyzer under potential control. All potentials were referred to the

372

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

4. Experimental results

copper deposition, whereas at higher negative potentials they had an activating influence on copper deposition. In addition, for sufficiently high negative potentials, the current increased regularly with the rotation speed, with or without chloride ions in the plating bath, showing that mass transport is the rate-limiting step for copper deposition.

4.1. Current–voltage curves

4.2. Electrochemical impedances

The current–voltage curves were plotted at various rotation velocities of the rotating disc electrode. Fig. 3 shows the curves obtained without (Fig. 3(A)) and with (Fig. 3(B)) chlorides in the copper bath. At low negative potentials, corresponding to low current densities, chloride ions had a rather inhibiting behaviour for

Fig. 4 shows the impedance measured at various rotation velocities of the rotating electrode in a solution containing chlorides. A diffusion loop is clearly shown at the lowest rotation velocities. This loop decreased as the rotation velocity increased. This supports the idea of mass transport limitation of the copper deposition

saturated sulphate electrode (SSE) and the temperature was maintained at 25 °C. The measurements were carried out in oxygen-free solutions by bubbling nitrogen.

0

-10

j / mA cm

-2

-20

-30 ρ = 40 rpm ρ = 100 rpm ρ = 400 rpm ρ = 2000 rpm

-40

-50

-60 -0.60

-0.58

-0.56

-0.54

-0.52

-0.50

-0.48

-0.46

-0.44

-0.42

-0.40

-0.44

-0.42

-0.40

E / V vs Hg/Hg2SO4

(A) 0

-10

j / mA cm

-2

-20

-30

ρ = 40 rpm

-40

ρ = 100 rpm ρ = 400 rpm ρ = 2000 rpm

-50

-60

-70 -0.60

(B)

-0.58

-0.56

-0.54

-0.52

-0.50

-0.48

-0.46

E / V vs Hg/Hg2SO4

Fig. 3. Experimental current–voltage curves plotted at various velocities of the rotating copper disc electrode. (A) Without chlorides, (B) with 103 M chlorides.

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

process. In the higher frequency range, a capacitive loop corresponding to the charge transfer resistance in parallel with a double layer capacitance of about 13 lF cm2 is observed. Between these two capacitive loops a third capacitive loop appears. Finally, an inductive loop is observed in the lowest frequency range. To demonstrate the presence of the third capacitive loop ascribed to the chlorides, Fig. 5 shows the comparison between the electrochemical impedances measured with and without chloride ions in the bath solution at various rotation velocities of the electrode. This comparison shows the presence of a capacitive loop in the intermediate frequency range, between the high frequency capacitive loop and the diffusion loop when chloride ions are added to the copper bath. As the diffusion loop was observed both in the presence and in the absence of chlorides, clearly it is the cupric ions, which are diffusing. Fig. 6 shows electrochemical impedances measured at a high rotation velocity (2000 rpm), for which mass transport was not the rate-limiting step, at various current densities in the presence of chlorides. Of course, the

-Imaginary Part Z(ω) / Ω cm

2

4 400 rpm ; 100 rpm 60 rpm ; 40 rpm 20 rpm

3

2

1

10000

0.1

10 1

0.01

1000 100

0

0.001 -1 0

1

2

3

Real Part Z(ω) / Ω cm

4

5

2

Fig. 4. Experimental electrochemical impedances measured at a 25 mA cm2 current density and at various rotation velocities of the rotating copper disc electrode in a solution containing 103 M NaCl.

[NaCl] = 10 -3M 40 rpm

40 rpm 2

-Im(∆E/∆I)/ Ω.cm

-Im(∆E/∆I)/Ω.cm

2

[NaCl] =0 1 4

0 3

-1 2

0

1

2

3

4

1 4

0 3

-1

5

0

Re(∆E/∆I)/Ω.cm

3 -1 2

3

4

5

100 rpm

0

3

-1 0

1

2

2

3

4

5

2

Re(∆E/∆I)/Ω.cm

-Im(∆E/∆I)/Ω.cm

2

2

2

-Im(∆E/∆I)/Ω.cm

4

4

5

Re(∆E/∆I)/Ω.cm

2

3 2

1

2

-Im(∆E/∆I)/ Ω.cm

2

4 0

1

2

Re(∆E/∆I)/ Ω.cm

100 rpm

1

-Im(∆E/∆I)/Ω.cm

1

2

2

0

373

400rpm

1

0

-1

400 rpm

1

0

-1

0

1

2

3

4

5

0

2

1

2

3

4

5

2

Re( ∆E/∆I)/Ω.cm

Re(∆E/∆I)/Ω.cm 3

Fig. 5. Comparison between the electrochemical impedances measured with 10 NaCl and without chloride in the bath solution at a 25 mA cm2 current density and at various rotation velocities of the rotating copper disc electrode. Measurements were carried out from 65 kHz down to 0.01 (or sometimes 0.001) Hz, decades are noted by black dots.

374

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375 2,0

jimp = 5 mA/cm

6 4

2

jimp = 25 mA/cm

2

jimp = 50 mA/cm

2

1000 100

2

10 1

0

2

2

8

0.001

-Imaginary Part Z( ω) / Ω cm

-Imaginary Part Z( ω) / Ω cm

2

10

-2

2

jimp = 25 mA/cm . 2

j imp = 50 mA/cm .

10 k

1,0

10 k 0,5

1k

10 k 1k

1

100 0.01

0,0

0.1

0.01

-0,5

0

(A)

j imp = 5 mA/cm .

1,5

2

4

6

8

Real Part Z (ω) / Ω cm

10

12

2

0,0

0,5

1,0

1,5

2,0

Real Part Z (ω) / Ω cm

(B)

2,5

2

Fig. 6. Electrochemical impedances measured at 2000 rpm at various current densities in a solution containing 0.75  103 M NaCl. (A) 5 mA cm2 , (B) 25 and 50 mA cm2 .

The current–voltage curves have shown that, at lower current densities, the chloride ions adsorb on the copper electrode to form CuCl, which partially blocks the electrode surface giving an inhibiting behaviour to chlorides. At higher current densities, the second step of the CuCl current path is faster and this path becomes an efficient parallel path for copper deposition. The currents I1 and I2 calculated from the model (Fig. 1) show that at low current density I1 is the major part of the total current whereas at higher current density it is I2 . Therefore the second path via CuCl is predominant for copper deposition. In addition, the model shows that at negative potentials the surface coverage h1 is predominant, which supports the major role of the CuCl deposition path (Fig. 7). The surface coverage h2 is predominant only at lower negative potentials. The general shape of the experimental electrochemical impedance is in agreement with the impedance calculated from the model under or without mass transport control (Fig. 2). However, to check in greater depth the validity of the model, the Rt I product was plotted with respect to the current density in Fig. 8. This product increased from 42 mV for the lower currents up to 52 mV for the highest currents. This shows a change of the predominant current path in the copper deposition

Surface coverages θ1 . θ2

5. Discussion

1,0

0,8

θ1 θ2

0,6

0,4

0,2

0,0 -0,6

-0,5

-0,4

-0,3

-0,2

Potential / V Fig. 7. Calculated variation of the surface coverages of CuCl, h1 , and anions A , h2 , with respect to the potential for the same values of the parameters given in Fig. 1.

60 58 -4

[NaCl] = 2.5.10 M

56

-4

[NaCl] = 5.10 M -4

54

[NaCl] = 7.5.10 M -3

[NaCl] = 10 M

52

R tc.I / mV

charge transfer resistance decreased when the current increased. The diameter of the capacitive loop characteristic of the presence of the chloride ions did not change greatly; only the characteristic frequency of this loop changes, particularly at high current density. The inductive loop is always present whatever the deposition current density.

50 48 46 44 42 40 38 0

10

20

30

j / mA cm

40

50

60

-2

Fig. 8. Variation of the product Rt I with respect to the current density for various chloride concentrations.

C. Gabrielli et al. / Journal of Electroanalytical Chemistry 572 (2004) 367–375

cies, e.g., anions, or to a nucleation process on the facets of the copper crystallites, which, both give rise to the relaxation of the surface area with potential. However, the real origin of the inductive feature is difficult to assess.

50 -2

j = 5 mA cm

Characteristic frequency / Hz

-2

40

375

j = 25 mA cm

-2

j = 50 mA cm

30

20

6. Conclusion

10

In this paper, we have tested a model where the chloride ions added to a sulphuric acid–sulfate copper plating bath have a catalytic action. As they are not consumed at the electrode, their action is not limited by mass transport and they are efficient even at low concentrations. The observed mass transport limitation is due to the diffusion of cupric ions.

0 0,00

0,25

0,50

0,75

1,00

-1

[NaCl] / mmol L

Fig. 9. Variation of the characteristic frequency of the capacitive loop relative to the chloride addition with respect to the chloride concentration at various current densities.

mechanism. This is qualitatively in agreement with the product Rt I calculated from the model (Fig. 1(B)), which increases with the overvoltage. Fig. 9 shows the characteristic frequency, i.e., the frequency of the largest magnitude of the imaginary part, of the capacitive loop observed in the presence of various concentrations of chloride ions taken from the data of Fig. 6 and similar experiments obtained at a high velocity (2000 rpm) of the rotating disc. The value of this characteristic frequency, fc , is equal to k3 c0 þ k4 þ k3 þ k4 fc ¼ 2pb1

ð52Þ

deduced from h1 ðxÞ in the model, i.e., for xb2  k5 þ k5 . It increased with the chloride ion concentration through k3 at high current density (50 mA cm2 ) showing that the chlorides have a major role at high overvoltages. In this current range the product Rt I increases showing the change of the prevailing mechanism. As predicted by the model, an inductive loop is observed in the lowest frequency range. This feature may be due to the adsorption–desorption of inhibiting spe-

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

[12] [13] [14] [15] [16]

E. Mattson, J. OÕM Bockris, Trans. Faraday Soc. 55 (1959) 1586. J. OÕM Bockris, M. Enyo, Trans. Faraday Soc. 58 (1962) 1187. O.R. Brown, H.R. Thirsk, Electrochim. Acta 10 (1965) 383. F. Chao, M. Costa, Bull. Soc. Chim. Fr. 10 (1968) 4015. I. Epelboin, F. Lenoir, R. Wiart, J. Cryst. Growth 13 (1972) 417. I.R. Burrows, J.A. Harrison, J. Thomson, J. Electroanal. Chem. 58 (1975) 241. G. Razumney, J. OÕM Bockris, J. Electroanal. Chem. 46 (1973) 185. J.A. Harrison, H.R. Thirsk, in: A.J. Bard (Ed.), Electroanalytical Chemistry, vol. 5, Marcel Dekker, New York, 1971, p. 67. E. Chassaing, R. Wiart, Electrochim. Acta 29 (1984) 649. P.C. Andricacos, C. Uzoh, J.O. Dukovic, J. Horkans, H. Deligianni, IBM J. Res. Develop. 42 (1998) 567. T.P. Moffat, J.E. Bonevich, W.H. Huber, A. Stanishevsky, D.R. Kelly, G.R. Stafford, D. Josell, J. Electrochem. Soc. 147 (2000) 4524. G.G. Lang, M. Ujvari, G. Horanyi, J. Electroanal. Chem. 522 (2002) 179. D.M. Soares, S. Wasle, K.G. Weil, K. Doblhofer, J. Electroanal. Chem. 532 (2002) 353. Z. Nagy, J.B. Blaudeau, N.C. Hung, L.A. Curtiss, D.J. Zurawski, J. Electrochem. Soc. 142 (1995) L87. M. Kang, A.A. Gewirth, J. Electrochem. Soc. 150 (2003) C426. A. Vicenzo, P.L. Cavallotti, J. Appl. Electrochem. 32 (2002) 743.