Ac

Na

b

c

a

ARRAA

KMCSTA

1

itedibuPaym

i

0h

Electrochimica Acta 109 (2013) 226– 232

Contents lists available at ScienceDirect

Electrochimica Acta

jou rn al hom ep age: www.elsev ier .com/ locate /e lec tac ta

n effective triblock copolymer as a suppressor for microvia filling viaopper electrodeposition

ing Xiaoa, Deyu Lia, Guofeng Cuib, Ning Lia,∗, Dong Tiana, Qing Li c, Gang Wuc,∗∗

Department of Applied Chemistry, Harbin Institute of Technology, Harbin 150001, ChinaSchool of Chemistry and Chemical Engineering, Sun Yat-sen University, Guangzhou 510275, ChinaMaterials Physics and Applications Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA

r t i c l e i n f o

rticle history:eceived 14 April 2013eceived in revised form 16 July 2013ccepted 16 July 2013vailable online xxx

eywords:icrovia filling

opper electroplatinguppressorriblock copolymer

a b s t r a c t

In this work, an effective suppressor was found in copper electroplating solution for microvia filling. Thesuppressor is a triblock copolymer comprised of ethylene oxide (EO)–propylene oxide (PO)–ethyleneoxide (EO) with a molecular weight of 2900 (EPE 2900). In terms of the obtained filling performanceevaluated by observing the obtained cross-sectional views of the microvias after electroplating, the elec-troplating solution was systematically studied as functions of Cl− concentration, EPE 2900 concentration,and convection conditions. It was found that filling performance was greatly dependent on these key fac-tors. In the developed copper electroplating formulation, EPE 2900 was able to yield sufficient fillingperformance in a broad Cl− concentration ranging from 3 to 200 ppm, although the optimized value wasdetermined to be about 30 ppm in terms of the best filling performance. During the microvia filling pro-cess, Cl− concentration, EPE 2900 concentration, and forced convection would affect each other. Their

−

dsorbing mechanism interactions were studied based on the anchor effect of Cl on the adsorption of EPE 2900 on Cu sub-strates. In doing so, extensive electrochemical experiments including galvanostatic measurements, linearsweep voltammetry, electrochemical impedance spectroscopy and cyclic voltammetry were employedto study the adsorption behavior of EPE 2900 on the cathode during both the copper electrodepositionand dissolution processes. An adsorbing mechanism of EPE 2900 on cathode was proposed to elucidatethe promotional role in improving the microvia filling performance.

© 2013 Published by Elsevier Ltd.

. Introduction

In electronic industry, microvia filling via copper electroplat-ng can be used to connect the adjacent two or more layers ofhe printed circuit board (PCB). In order to achieve the reliablelectrical interconnection, the microvia should be fully filled byepositing copper without seam or void in an acid copper plat-

ng solution containing different key additives [1–8]. As reportedefore, “superfilling” or “bottom-up filling” can be realized bysing polyethylene glycol (PEG) as suppressor, such as PEG–Cl,EG–SPS–Cl, PEG–MPS–Cl, PEG–SPS–Cl–JGB, etc. [9–18]. However,

new class of suppressor using triblock copolymers (EPE) of eth-lene oxide (EO)–propylene oxide (PO)–ethylene oxide (EO) for

icrovia filling is not extensively reported yet [19–21].In our previous reports [22,23], it was demonstrated that there

s an obvious synergistic effect between EPE 2900 (an EPE triblock

∗ Corresponding author. Tel.: +86 451 86413721; fax: +86 0451 86413721.∗∗ Corresponding author. Tel.: +1 505 665 0659.

E-mail addresses: lininghit@263.net (N. Li), wugang@lanl.gov (G. Wu).

013-4686/$ – see front matter © 2013 Published by Elsevier Ltd.ttp://dx.doi.org/10.1016/j.electacta.2013.07.127

copolymer with a molecular weight of 2900) and Cl− on inhibitingcopper deposition during the microvia filling process. Given a fixedEPE 2900 concentration, there existed an optimal Cl− concentra-tion to achieve the maximum synergistic effect. The suppressionstrength and microvia filling performance obtained from such typeof triblock copolymers of EO–PO–EO with various molecular weightand EO content were systematically compared [23]. Among thestudied triblock copolymers, the EPE 2900 exhibited the strongestsuppression strength on inhibiting copper deposition, showing thebest performance during the microvia filling. However, the adsorp-tion behavior of EPE 2900 on the cathode was not elucidated yet.Further understanding its inhibiting mechanism will be greatlyhelpful to guide the selection of suppressor in microvia filling viacopper electroplating.

In this work, microvia filling performance of the copper platingsolution using EPE 2900 as a suppressor is investigated by means ofextensive electrochemical methods. It has been found that the EPE

2900 can form micelles with a hydrophilic shell (PEO segments) anda hydrophobic core (PPO segments) in aqueous solution [24,25].The correlations between obtained filling performance and stud-ied solution formulations indicate that the copper plating solution

ica Ac

ciwsiiaitbtta

2

2

uda1wppoip1umf

mbwtocsiu2

N. Xiao et al. / Electrochim

ontaining the EPE 2900 with an optimal concentration (200 ppm)s able to offer sufficient filling performance in a wide operation

indow of Cl−. Specially, with the assistance of forced convection,uperfilling can be fully achieved in the plating solution contain-ng Cl− as low as 3 ppm. Interestingly, when Cl− concentration isncreased to 30 ppm, the role of EPE 2900 is overwhelmed by Cl−,nd further increasing the triblock copolymer concentration showsnsignificant influence on the filling performance. A series of elec-rochemical measurements are carried out to study the adsorptionehavior of the EPE 2900 on the cathode during the copper elec-rodeposition, with an aim to establish the correlations betweenhe microvia filling performance and the concentration of EPE 2900nd Cl−.

. Experimental

.1. Microvia filling via copper electroplating

PCBs with microvias fabricated by CO2 laser ablation weresed as testing samples for the microvia filling experiments. Theimension of the PCB sample was 5 × 12 cm2 with an effectiverea of 5 × 9 cm2. The diameter and depth of the microvia were50 �m and 75 �m, respectively. Prior to microvia filling, the side-alls of these microvias were metallized by electroless copperlating to deposit a 2–3 �m copper film. In these experiments,hosphorus-containing (0.03–0.065 wt%) copper plates with a sizef 6 × 12 cm2 were used as anodes and directly placed in the plat-ng bath containing 1.5 L solution. The current density for copperlating was controlled at 2 A/dm2 and the electroplating time was00 min. Unless otherwise specified, forced convection by contin-ously bubbling air flow in the solution was applied during theseicrovia filling experiments so as to ensure efficient mass trans-

er.The electroplating solution used for microvia filling experi-

ents was composed of 220 g/L CuSO4·5H2O, 55 g/L H2SO4, 3 ppmis(3-sulfopropyl) disulfide (SPS), 4 ppm Janus Green B (JGB) asell as Cl− (added as NaCl), EPE 2900 with varying concentra-

ion. EPE 2900 (BASF) was a triblock copolymer with the structuref EO–PO–EO. It has a nominal molecular weight of 2900 andontains approximately 40 wt% EO. Temperature of the plating

olution was controlled at 25 ◦C. The filling performance of the plat-ng bath was evaluated by cross-sectional views of the microviassing optical microscopy (DFC290, Leica) at a magnification of00×.

Fig. 1. (a1–a6) Cross-sectional views of microvias after electroplating in the plati

ta 109 (2013) 226– 232 227

2.2. Electrochemical measurements

Galvanostatic measurements (GMs), linear sweep voltammetry(LSV), and electrochemical impedance spectroscopy (EIS) wereperformed on an electrochemical workstation (CHI 760D) in athree-electrode cell at 25 ◦C. A platinum rotating disk electrode(Pt-RDE, Pine) with a diameter of 5 mm and a small piece of cop-per bar were employed as the working electrode and counterelectrode, respectively. A saturated mercury–mercurous sulfateelectrode (MSE) was used as the reference electrode. Prior to elec-trochemical measurements, a thin copper layer with a thickness of500 nm was deposited onto the Pt-RDE in the base electrolyte toprepare a Cu-RDE.

The GMs were conducted at a constant current density of2 A/dm2 and at different rotating speeds of 100, 400, and 1000 rpm.According to the reported literatures [11,26], �� values, whichcorresponded to an averaged potential difference measured at arotating speeds of 100 and 1000 rpm, can be used as indicators ofthe filling performance when forced convection was applied in theplating bath. Larger �� values usually suggest better filling per-formance. In the meantime, the �� value between 100 rpm and400 rpm was calculated to evaluate the filling performance of theplating solution without forced convection.

The LSV polarization plots were recorded with a negative-goingsweep from open circuit potential to −0.7 V vs. MSE at a scan rateof 1 mV/s. EIS measurements were performed at −0.65 V vs. MSE.The frequency range was from 100 K to 0.1 Hz with an amplitude of5 mV. The impedance data were fitted using equivalent circuit byusing ZSimpWin software. The rotating speed of the Cu-RDE washeld at 1000 rpm during the LSV and EIS measurements.

Cyclic voltammetry (CV) was carried out using a Pt/Pt rotatingring-disk electrode (RRDE) as the working electrode (Pt disk radiusr1 = 4.57 mm; Pt ring inner radius r2 = 4.93 mm and outer radiusr3 = 5.38 mm). The Pt/Pt RRDE was mechanically polished with alu-mina powders before each measurement. The counter electrodewas a Pt plate with an area of 1 cm2 and the reference electrodewas MSE. The potential of the disk was negatively swept from 1.0 Vto −0.6 V and then back to 1.0 V vs. MSE with a scan rate of 5 mV/s.In the meantime, the ring potential was kept at 0.2 V vs. MSE whichwas positive enough to oxidize the Cu+ to Cu2+ at the ring elec-

trode. The rotation speed of the RRDE was 1000 rpm during the CVmeasurements.

Unless otherwise specified, electrolyte used for all of the elec-trochemical measurements was composed of 220 g/L CuSO4·5H2O,

ng solution containing 200 ppm EPE 2900 and different Cl− concentration.

228 N. Xiao et al. / Electrochimica Ac

0 50 0 100 0 150 0

-0.66

-0.55

-0.50

30ppm-1000 rpm90ppm-1000 rpm200pp m-1000 rpm

200ppm-100 rpm

90ppm-100rpm

60pp m-1000 rpm

60ppm-100rpm

30ppm-100 rpm

3ppm-1000 rpm

3ppm-100 rpm

0ppm-1000 rpm

0ppm-100 rpm

E v

s M

SE

/V

Plating t ime/s

Fig. 2. GMs performed at two different rotating speeds in the plating solution con-taining 200 ppm EPE 2900 and different Cl− concentration.

Table 1Potential differences calculated from GMs shown in Fig. 2.

5ti

3

3

irfdt1ttw

FC

Concentration of Cl− (ppm) 0 3 30 60 90 200Potential Differences �� (mV) –10.1 14.2 19.5 16.2 11.9 5.0

5 g/L H2SO4, 3 ppm SPS, 4 ppm JGB, Cl− and EPE 2900, in whichhe concentration of EPE 2900 and Cl− was specified in each exper-ment.

. Results and discussion

.1. Role of Cl− in microvia filling

The cross-sectional views of the microvias and the correspond-ng GMs as a function of Cl− concentration are shown in Figs. 1 and 2,espectively. Without adding any Cl− in the plating solution, con-ormal deposition is observed in the microvia (Fig. 1(a)), which isetrimental to PCB fabrication. As displayed in Fig. 2, the poten-ial obtained at 1000 rpm is more negative than that obtained at

00 rpm with a �� value of −10.1 mV listed in Table 1. Accordingo the galvanostat method proposed by Dow et al. [11,27], the nega-ive �� value implies that no superfilling can occur in the microvia,hich is in good agreement with our observed microvia filling

ig. 3. Cross-sectional views of microvias after electroplating in the plating solution contal− .

ta 109 (2013) 226– 232

results. After 3 ppm Cl− is added into the plating bath, bottom-up deposition appears and the microvias are almost fully filled bydeposited copper. Only a small dimple is observed near the mouthof the microvia. When the Cl− concentration is further increasedup to 200 ppm, the size of the dimple firstly decreases and thenincreases. As a result, the plating solution yields the best fillingperformance at a Cl− concentration of 30 ppm. This is in good agree-ment with the GM results, which the �� value obtained in theplating solution containing 30 ppm Cl− is the greatest, as listed inTable 1. The obtained results indicate that galvanostat method iseffective and reliable to evaluate the microvia filling performancein our experiments.

3.2. Role of EPE 2900 in microvia filling

At a very low Cl− concentration of 3 ppm, the filling performanceis greatly dependent on the concentration of EPE 2900, as shown inFig. 3(a1)–(a4). When EPE 2900 concentration is increased from 10to 50 ppm, a large dimple near the mouth of the microvia can still beobserved. However, upon the concentration reaching 100 ppm, thedimple almost disappears. This phenomenon is due to enhancedsuppression strength on copper surface caused by electrostaticadsorption of EPE 2900. Further increasing EPE 2900 concentra-tion to 200 ppm, the dimple in the microvia becomes larger againas shown in Fig. 3(a4), which can be attributed to the insignificantdifference of suppression between inside the microvia and on theboard surface.

It is worth noting that the filling performance measured withthe plating solution containing a high Cl− concentration (30 ppm)becomes less dependent on the EPE 2900 concentration. Themicrovia can be well filled in the plating solution containing only10 ppm EPE 2900 (Fig. 3(b1)). Since Cl− is able to serve as anchorsfor the subsequent adsorption of the suppressor [28], the adsorp-tion of EPE 2900 on the copper surface could be greatly enhanced inthe presence of higher Cl− concentration. Thus, the copper surfaceis totally covered by EPE 2900 even at its relatively low concentra-tion. As a result, an increase of EPE 2900 cannot significantly furtherimprove filling performance, that is, the filling performance seemsto be independent of EPE 2900 concentration.

3.3. Role of forced convection in microvia filling

According to the convection-dependent adsorption (CDA)mechanism [28], strong convection can selectively transfer more

ining different EPE 2900 and Cl− concentration: (a1–a4) 3 ppm Cl−; (b1–b4) 30 ppm

N. Xiao et al. / Electrochimica Ac

0 50 0 100 0 150 0

-0.70

-0.65

-0.60

-0.55

a

b

200 pp m Cl-

60 ppm Cl-

10 ppm Cl-

3 pp m Cl-

E v

s M

SE

/V

1000 rpm400 rpm

Plating time/s

100 rpm

0 500 100 0 150 0-0.60

-0.58

-0.56

-0.54

-0.52

E v

s M

SE

/V

=7 mV

=0 mV

(a)

(c)

(b)

Plating time/s

(a) at 100 rpm

(b) at 400 rpm

(c) at 1000 rpm

Fig. 4. (a) GMs performed consecutively at three different rotating speeds in theplating solution containing 200 ppm EPE 2900 and different Cl− concentration; (b)GMs performed separately in the plating solution containing 200 ppm EPE 2900 and3 ppm Cl− .

Fig. 5. Cross-sectional views of microvias after electroplating from the plating solutionwithout (b1–b4) forced convection.

ta 109 (2013) 226– 232 229

Cl− ions to the cathode surface per unit time, thereby facilitat-ing the subsequent adsorption of EPE 2900 in the plating solution.Thus, the deposition rate of copper at outside microvias can bedecreased when more Cl− ions are around due to the enhanced con-vection strength. As displayed in Fig. 4(a), three curves measuredwith the plating solution correspondingly containing 10, 60, and200 ppm Cl− attest a great dependence of microvia performanceon the strength of forced convection. Hence, the absolute valuesof the potential gradually increase with the rotating speed varyingfrom 100 to 1000 rpm. This is in good agreement with the proposedCDA mechanism [28].

As a matter of fact, the potential differences between 100 and400 rpm (��100 rpm–400 rpm) and those between 100 and 1000 rpm(��100 rpm–1000 rpm) are both positive, which indicates that super-filling can be realized in these plating solutions regardless ofwhether forced convection is applied or not. However, the curvemeasured with 3 ppm Cl− is found to be somehow independent ofrotating speed ranging from 100 to 400 rpm. The recorded curvesare unstable during the measurements, which make it difficult todetermine the �� values. Thus, GMs was performed in the platingsolution containing 3 ppm Cl− at different rotating speeds rangingfrom 100 to 1000 rpm. As shown in Fig. 4(b), while the �� valuebetween 100 and 400 rpm is about 0 mV, this value is increasedto 7 mV between 100 and 1000 rpm. This indicates that superfill-ing cannot effectively occur with the plating solution containing3 ppm Cl− unless a strong forced convection is applied. In addition,the results obtained from Fig. 4 are in good agreement with themicrovia filling experiments exhibited in Fig. 5. Specifically, only isthe Cl− concentration higher than 10 ppm, its filling performancebecomes independent of forced convection.

3.4. Adsorption behavior of EPE 2900

In order to further understand the suppression role of EPE 2900during the microvia filling process and explain the results shown inFig. 3, extensive electrochemical measurements are performed inthe plating solution with different Cl− and EPE 2900 concentrationto study the adsorption behavior of EPE 2900.

In these experiments, the plating solution containing 10 ppm

EPE 2900 and 3 ppm Cl− is used as a base formula, further addi-tion of EPE 2900 or Cl− is carried out to study their influences oncopper electrodeposition. As shown in Fig. 6, increased overpo-tential of copper deposition can be observed at a current density

containing 200 ppm EPE 2900 and different Cl− concentration: with (a1–a4) and

230 N. Xiao et al. / Electrochimica Acta 109 (2013) 226– 232

-0.7 -0.6 -0.5 -0.4

-3

-2

-1

0

e d c b a

Curr

ent d

ensity

/A d

m-2

E vs MSE /V

Fig. 6. The current–potential curves measured by LSV in the plating solution con-t −

C1

ocpEbtscE

optawPawlt

Fc

0 50 100 150 200 250

0

50

100

150

200

250

-Z

"/⏐

cm

2

Z'/⏐ cm2

10 ppm EPE 2900 + 3 ppm Cl-

100 ppm EPE2900 + 3 ppm Cl-

10 ppm EPE 2900 + 30 ppm Cl-

100 ppm EPE2900 + 30 ppm Cl-

RS

R1

CPE1 CPE2

R2 L

(b)

(a)

Fig. 8. (a) Nyquist plots recorded from the plating solution containing different

aining different EPE 2900 and Cl concentration: (a) 0 ppm EPE 2900 and 3 ppml−; (b) 10 ppm EPE 2900 and 3 ppm Cl−; (c) 100 ppm EPE 2900 and 3 ppm Cl−; (d)0 ppm EPE 2900 and 30 ppm Cl−; (e) 100 ppm EPE 2900 and 30 ppm Cl− .

f 2 A/dm2. In other words, with the increase of EPE 2900 or Cl−

oncentration, the suppression strength of EPE 2900 toward cop-er deposition is enhanced. It suggests that there should be morePE 2900 molecules adsorbed onto the cathode surface. Due to thelockage resulted from adsorption of EPE 2900, the Cu2+ ions inhe plating solution become more difficult to diffuse to the cathodeurface. In addition, the active sites for copper deposition on theathode surface are decreased due to a large number of adsorbedPE 2900 molecules.

It should be noted that, compared to the EPE 2900, variationf Cl− concentration exhibit more significant effect on the over-otential of copper electrodeposition. This phenomenon is due tohe different adsorption behavior of EPE 2900 in the presence orbsence of Cl−. It is very likely that EPE 2900 is able to complexith Cu+ and Cl− to form EPE 2900–Cu+–Cl−, which is similar to

EG–Cu+–Cl− [7]. Because EPE 2900 and PEG are both polyethersnd contain the same structure units of EO, this complex can replace

ater molecules and strongly adsorb onto the inner Helmholtz

ayer of the cathode to inhibit copper deposition [7,29]. In con-rast, most EPE 2900 molecules only weakly adsorb onto the outer

0 200 400 600 80 0 100 0

-0.65

-0.60

-0.55

-0.50

10 ppm EPE 2900 + 3 ppm Cl-

100 ppm EPE2900 + 3 ppm Cl-

10 ppm EPE2900 + 30 ppm Cl-

100 ppm EP E29 00 + 30 pp m Cl-

E v

s M

SE

/V

Plating t ime/s

ig. 7. GMs performed in the plating solution containing different EPE 2900 and Cl−

oncentration.

EPE 2900 and Cl− concentrations. Point: tested data, solid line: fitted curve. (b)Equivalent circuit of EIS.

Helmholtz layer of the electrode via electrostatic and Van der Waalsforce when Cl− is absent or insufficient [29].

As shown in Fig. 7, the stable potential of copper electrodepo-sition shifts negatively when the concentration of either EPE 2900or Cl− increases. In particular, the increase of Cl− shows moresignificant influence on the stable potential compared to EPE 2900.All these results are consistent with LSV measurements.

In order to provide insights into the adsorption behavior ofEPE 2900 as a function of Cl−concentration, EIS measurementsare employed to study the copper deposition. The Nyquist plotsrecorded at −0.65 V vs. MSE for different plating solution arecompared in Fig. 8(a). The corresponding equivalent circuit ispresented in Fig. 8(b). It can be seen that the simulated plots rep-resented by solid lines are well fit with the experimental ones.According to the literatures [29–31], in the equivalent circuit, Rs

represents the resistance of the solution. The double layer capac-ity of CPE1 in parallel to the charge transfer resistance of R1is related to the capacitive loop in the high frequency. CPE2 inparallel to R2 corresponding to the second capacitive loop arecaused by the relaxation process of (Cuads

+). L corresponding to theinductive loop is resulted from the relaxation process of Cu+–SPS.It is well known that the electrodeposition of copper usuallyincludes two steps: Cu2+ + e− → Cu+ (step 1) and Cu+ + e− → Cu

(step 2) [32]. Among them, step 1 is a kinetically slow reac-tion and is considered to be the rate-determining step. In theemployed equivalent circuit, the size of R1 is used to evaluate the

Table 2Simulated impedance parameters as functions of EPE 2900 and Cl− concentration.

Additive concentration Rs/� cm2 R1/� cm2 R2/� cm2

10 ppm EPE 2900 + 3 ppm Cl− 5.4 66.7 0.1100 ppm EPE 2900 + 3 ppm Cl− 5.0 79.2 1.810 ppm EPE 2900 + 30 ppm Cl− 5.0 129.4 19.3100 ppm EPE 2900 + 30 ppm Cl− 4.8 227.2 21.3

N. Xiao et al. / Electrochimica Acta 109 (2013) 226– 232 231

-0.6 -0.4 -0.2 0.0

-6

-4

-2

0

2

4 a b

c d

(Idis

kx1

0-3) /

A

Edisk

vs MSE/V

0

1

2

(Irin

gx1

0-4)/

A

Fig. 9. CV plots measured using Pt/Pt RRDE in the plating solution containing dif-ferent EPE 2900 and Cl− concentration: (a) 10 ppm EPE 2900 and 3 ppm Cl−; (b)1 − −

E

sd

fffic6cfmtb

dttsppTpoAtct

dfHatdcoCCedoce

-0.6 -0.4 -0.2 0.0 0.2-20

-10

0

10

(Idis

k x

10

-3) /

A

Edisk

vs MSE /V

a b

c d

0

2

4

6

(Irin

g x

10

-4)/

A

Fig. 10. CV plots measured using Pt/Pt RRDE in the electrolytes: (a) base solution

sufficient filling performance in a wide range of Cl− concentra-

00 ppm EPE 2900 and 3 ppm Cl ; (c) 10 ppm EPE 2900 and 30 ppm Cl ; (d) 100 ppmPE 2900 and 30 ppm Cl− .

uppression strength of EPE 2900 and Cl− on inhibiting coppereposition.

As listed in Table 2, the corresponding R1 values are increasedrom 66.7 to 79.2 � cm2 when EPE 2900 concentration increasesrom 10 to 100 ppm at a constant Cl− concentration of 3 ppm. Uponxing the EPE 2900 concentration at 10 ppm, increasing Cl− con-entration from 3 to 30 ppm leads to an increase of R1 values from6.7 to 129.4 � cm2. Obviously, the rise either in EPE 2900 or Cl−

oncentration can enhance the suppression strength of EPE 2900or inhibiting copper deposition. It is worth noting that enhance-

ent of the suppression strength caused by Cl− is more significanthan that caused by EPE 2900, thereby attesting that adsorptionehavior of EPE 2900 is changed due to the presence of Cl−.

In order to provide more insights into the mechanism of coppereposition and obtain direct evidences of EPE 2900 for inhibi-ing copper deposition, rotating ring-disk electrode is employedo study the possible intermediate, Cu+, during both copper depo-ition and copper stripping processes. As shown in Fig. 9, in thelating solution containing 10 ppm EPE 2900 and 3 ppm Cl−, cop-er is deposited on the Pt disk at potentials below −0.4 V vs. MSE.he reverse scan of the voltammogram shows an anodic strippingeak at −0.24 V vs. MSE due to copper dissolution. Further additionf EPE 2900 or Cl− leads to negative shifts of deposition potentials.t the same time, the copper stripping peaks become smaller. All

hese results indicate that copper deposition becomes more diffi-ult on the disk with larger overpotentials, which is consistent withhe LSV, GMs and EIS results discussed above.

At the ring electrode, Cu+, the intermediate during the coppereposition, can be detected by re-oxidizing it to Cu2+. It can be seenrom Fig. 9 that the oxidation of Cu+ is observed at −0.42 V vs. MSE.owever, no Cu+ ion is detected at the ring electrode at more neg-tive potentials, where all of Cu+ ions should be reduced to Cu athe disk electrode. During copper dissolution process, the Cu+ isetected again at the ring since it is also an intermediate during theopper dissolution process. It can be seen from Fig. 9 that the peakf Cu+ detected at the ring becomes smaller after either EPE 2900 orl− is added into the plating solution. This suggests that EPE 2900 orl− can effectively prevent the Cu+ escaping from the disk to ringlectrode. Considering the depolarization effect of Cl− on coppereposition process, the most reasonable reason is that the addition

f EPE 2900 or Cl− can lead to formation of EPE 2900–Cu+–Cl− thatan be strongly adsorbed onto the cathode. As a result, the Cu+ gen-rated on the disk cannot be easily escaped and then detected by

without EPE 2900 and Cl−; (b) 30 ppm Cl−; (c) 100 ppm EPE 2900; (d) 30 ppm Cl−

and 100 ppm EPE 2900. The base solution was composed of 220 g/L CuSO4·5H2O and55 g/L H2SO4.

the ring electrode. These results are also in good agreement withthe EIS data.

In order to study the possible interaction between EPE 2900 andCu+, CV measurements are performed in the electrolytes containingvarious EPE 2900 and Cl− concentration. By using Pt/Pt RRDE, theCV plots are shown in Fig. 10, in accordance with those presentedin our previous work [22]. On the ring electrode, solely introduc-ing Cl− into the electrolyte can increase the Cu+ peak during copperdeposition. This can be attributed to the depolarization effect of Cl−,which can accelerate the rate determining step (Cu2+ + e− → Cu+)during the copper deposition [33,34]. On the contrary, the Cu+

peak during the copper deposition almost disappeared after exclu-sively adding EPE 2900, without Cl−. The reasonable explanationis that EPE 2900 adsorbed onto the disk electrode can react withthe intermediate Cu+ to form an EPE 2900–Cu+ complex. Interest-ingly, the Cu+ peak during the copper dissolution process is greatlyincreased as shown in Fig. 10(c), which is probably due to the for-mation of EPE 2900–Cu+ complex. Compared to Cu+, this complexis relatively stable and can exist in the electrolyte for a longer time,but it can no longer adsorb onto the disk electrode due to its posi-tive charge during the copper dissolution process. In the electrolytesimultaneously containing EPE 2900 and Cl−, the Cu+ peak detectedduring copper deposition process cannot be as obvious as observedbefore. The corresponding peak potentials shift negatively duringdissolution process due to a trend of formation of the Cu+ com-plex. On the disk electrode, the overpotential for copper depositionis greatly increased, and thus the corresponding copper strippingpeak decreases obviously. This is also due to the formation of EPE2900–Cu+–Cl− complex, strongly blocking the copper deposition.Compared to the complex of EPE 2900–Cu+, the EPE 2900–Cu+–Cl−

is even more stable in the electrolyte and has stronger adsorptionstrength on the electrode.

4. Conclusions

In this work, a triblock copolymer (EPE 2900) is systematicallystudied as an effective suppressor in microvia filling via copper elec-troplating. Both filling performance evaluation and galvanostaticmeasurements indicate that the copper plating solution contain-ing an optimal EPE 2900 concentration (200 ppm) is able to yield

tions. Forced convection applied during the deposition is ableto greatly reduce the required Cl− concentration to 3 ppm, stillretaining good microvia filling performance. Likewise, when the

2 ica Ac

Cpcd3caTfieeCc2occcaW

A

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[33] Z. Nagy, J.P. Blaudeau, N.C. Hung, L.A. Curtiss, D.J. Zurawski, Chloride ion catal-

32 N. Xiao et al. / Electrochim

l− concentration increases from 10 to 200 ppm, the fillingerformance becomes more independent of the applied forcedonvection. Furthermore, filling performance obtained from theeposition solution containing high Cl− concentration (above0 ppm) is also found to be almost independent of EPE 2900 con-entration, likely due to the anchoring effect of Cl− during thedsorption of EPE 2900 so as to reduce the usage of EPE 2900.hese findings can rationally guide to design an effective microvialling formula with minimum addition of additives. A series oflectrochemical measurements are performed to elucidate copperlectrodeposition in the microvias in the presence of EPE 2900 andl−, which well correlate with the filling performance. This effortan enhance the understanding of the adsorption behavior of EPE900 and its interaction with Cl− and intermediates Cu+ at the cath-de surface. Importantly, in the presence of sufficient Cl−, EPE 2900an react with Cu+ and Cl− to form a stable EPE 2900–Cu+–Cl−

omplex, capable of strongly adsorbing onto the cathode to inhibitopper deposition. On the contrary, insufficient Cl− leads to weakdsorption of EPE 2900 on the cathode via electrostatic and Van deraals force, exhibiting a low suppressing efficiency.

cknowledgement

This work was financially supported by Highnic Group (China).

eferences

[1] W.P. Dow, C.C. Li, Y.C. Su, S.P. Shen, C.C. Huang, C. Lee, B. Hsu, S. Hsu, Microviafilling by copper electroplating using diazine black as a leveler, Electrochim.Acta 54 (2009) 5894.

[2] T. Kobayashi, J. Kawasaki, K. Mihara, H. Honma, Via-filling using electroplatingfor build-up PCBs, Electrochim. Acta 47 (2001) 85.

[3] W.P. Dow, C.C. Li, M.W. Lin, G.W. Su, C.C. Huang, Copper fill of microvia usinga thiol-modified Cu seed layer and various levelers, J. Electrochem. Soc. 156(2009) D314.

[4] A.C. West, S. Mayer, J. Reid, A superfilling model that predicts bump formation,Electrochem. Solid-State Lett. 4 (2001) C50.

[5] S.-K. Kim, D. Josell, T.P. Moffat, Cationic surfactants for the control of overfillbumps in Cu superfilling, J. Electrochem. Soc. 153 (2006) C826.

[6] H.-C. Tsai, Y.-C. Chang, P.-W. Wu, Rapid galvanostatic determination on levelersfor superfilling in Cu electroplating, Electrochem. Solid-State Lett. 13 (2010) D7.

[7] Z.V. Feng, X. Li, A.A. Gewirth, Inhibition due to the interaction of polyethyleneglycol, chloride, and copper in plating baths: a surface-enhanced Raman study,J. Phys. Chem. B 107 (2003) 9415.

[8] T.P. Moffat, D. Wheeler, S.-K. Kim, D. Josell, Curvature enhanced adsorbate cov-erage mechanism for bottom-up superfilling and bump control in damasceneprocessing, Electrochim. Acta 53 (2007) 145.

[9] M. Hayase, M. Taketani, K. Aizawa, T. Hatsuzawa, K. Hayabusa, Copper bottom-up deposition by breakdown of PEG–Cl inhibition, Electrochem. Solid-StateLett. 5 (2002) C98.

10] M. Tan, C. Guymon, D.R. Wheeler, J.N. Harb, The role of SPS, MPSA, and chlo-

ride in additive systems for copper electrodeposition, J. Electrochem. Soc. 154(2007) D78.

11] W.P. Dow, C.W. Liu, Evaluating the filling performance of a copper platingformula using a simple galvanostat method, J. Electrochem. Soc. 153 (2006)C190.

[

ta 109 (2013) 226– 232

12] T.P. Moffat, D. Wheeler, D. Josell, Electrodeposition of copper in the SPS–PEG–Cladditive system, J. Electrochem. Soc. 151 (2004) C262.

13] Y. Cao, P. Taephaisitphongse, R. Chalupa, A.C. West, Three-additive model ofsuperfilling of copper, J. Electrochem. Soc. 148 (2001) C466.

14] I. Volov, A.C. West, Interaction between SPS and MPS in the presence of ferrousand ferric ions, J. Electrochem. Soc. 158 (2011) D456.

15] T.P. Moffat, L.-Y.O. Yang, Accelerator surface phase associated with supercon-formal Cu electrodeposition, J. Electrochem. Soc. 157 (2010) D228.

16] M.L. Walker, L.J. Richter, T.P. Moffat, Potential dependence of competitiveadsorption of PEG, Cl− , and SPS/MPS on Cu, J. Electrochem. Soc. 154 (2007)D277.

17] M. Tan, J.N. Harb, Additive behavior during copper electrodeposition in solu-tions containing Cl− , PEG, and SPS, J. Electrochem. Soc. 150 (2003) C420.

18] M.E.H. Garrido, M.D. Pritzker, Voltammetric study of the inhibition effect ofpolyethylene glycol and chloride ions on copper deposition, J. Electrochem.Soc. 155 (2008) D332.

19] J.W. Gallaway, M.J. Willey, A.C. West, Copper filling of 100 nm trenches usingPEG, PPG, and a triblock copolymer as plating suppressors, J. Electrochem. Soc.156 (2009) D287.

20] J.W. Gallaway, M.J. Willey, A.C. West, Acceleration kinetics of PEG, PPG, and atriblock copolymer by SPS during copper electroplating, J. Electrochem. Soc.156 (2009) D146.

21] A. Survila, Z. Mockus, S. Kanapeckaite, Effect of halides on adsorption propertiesof polyether laprol 2402 C on copper and tin electrodes, J. Electroanal. Chem.552 (2003) 97.

22] N. Xiao, N. Li, D.Y. Li, D. Tian, R.Q. Liu, G. Wu, The synergistic effect betweena triblock copolymer and chloride ions in Cu electrodeposition into microvias,ECS Solid-State Lett. 1 (2012) P1.

23] N. Xiao, N. Li, G.F. Cui, D. Tian, S.Y. Yu, Q. Li, G. Wu, Triblock copolymers assuppressors for microvia filling via copper electroplating, J. Electrochem. Soc.160 (2013) D1.

24] P.N. Hurter, T.A. Hatton, Solubilization of polycyclic aromatic hydrocarbonsby poly(ethylene oxide-propylene oxide) block copolymer micelles: effects ofpolymer structure, Langmuir 8 (1992) 1291.

25] L. Zhou, S. Schlick, Electron spin resonance (ESR) spectra of amphiphilicspin probes in the triblock copolymer EO13PO30EO13 (pluronic L64): hydra-tion, dynamics and order in the polymer aggregates, Polymer 41 (2000)4679.

26] W.P. Dow, M.Y. Yen, C.W. Chou, C.W. Liu, W.H. Yang, C.H. Chen, Practical moni-toring of filling performance in a copper plating bath, Electrochem. Solid-StateLett. 9 (2006) C134.

27] W.P. Dow, M.Y. Yen, W.B. Lin, S.W. Ho, Influence of molecular weight ofpolyethylene glycol on microvia filling by copper electroplating, J. Electrochem.Soc. 152 (2005) C769.

28] W.P. Dow, M.Y. Yen, S.Z. Liao, Y.D. Chiu, H.C. Huang, Filling mechanism inmicrovia metallization by copper electroplating, Electrochim. Acta 53 (2008)8228.

29] W. Wang, Y.B. Li, Effect of Cl− on the adsorption–desorption behavior of PEG, J.Electrochem. Soc. 155 (2008) D263.

30] S. Varvara, L. Muresan, I.C. Popescu, G. Maurin, Kinetics of copper elec-trodeposition in the presence of triethyl-benzyl ammonium chloride, J. Appl.Electrochem. 33 (2003) 685.

31] C. Gabrielli, P. Moc otéguy, H. Perrot, D. Nieto-Sanz, A. Zdunek, A model forcopper deposition in the damascene process, Electrochim. Acta 51 (2006)1462.

32] E. Mattsson, J.O’M. Bockris, Galvanostatic studies of the kinetics of depositionand dissolution in the copper + copper sulphate system, Trans. Faraday Soc. 55(1959) 1586.

ysis of the copper deposition reaction, J. Electrochem. Soc. 142 (1995) L87.34] W.P. Dow, H.S. Huang, Roles of chloride ion in microvia filling by copper elec-

trodeposition. I. Studies using SEM and optical microscope, J. Electrochem. Soc.152 (2005) C67.