20 effect of plating mode, thiourea and chloride on the morphology of.pdf

8/10/2019 20 Effect of plating mode, thiourea and chloride on the morphology of.pdf

1/13

Electrochimica Acta 50 (2005) 18491861

Effect of plating mode, thiourea and chloride on the morphology of copper deposits produced in acidic sulphate solutions

Nisit Tantavichet 1, Mark D. Pritzker Department of Chemical Engineering, University of Waterloo, Waterloo, Ont., Canada N2L 3G1

Received 14 May 2004; received in revised form 30 July 2004; accepted 14 August 2004

Abstract

The inuence of plating mode, chloride and thiourea (TU) on morphology of copper deposits has been studied. All experiments wereconducted on disc electrodes rotating at 500 rpm and an average current density of 4A dm 2 to produce 10 m thick deposits. In additive-freesolutions, theuse of pulsed current (PC) improveddeposit morphology andbrightness over DCplating. In thepresenceof thiourea (noCl ),thedepositsobtainedby DCandPC plating weresimilarunder mostplatingconditions. Thepresenceof thioureagenerally improveddepositqualityover thatobtainedin additive-free solutions,but caused theformationof microscopicnodules and thedeposits to appearslightly cloudy, resultingin lower reectances than that of a polished uncoated copper surface. The addition of Cl to thiourea-containing solutions strongly inuenceddeposit morphology at both microscopic and macroscopic scales depending on chloride concentration and pulse conditions. It preventednodule formation and created microscopically bright and reective deposits, but caused extreme macroscopic roughness. Nevertheless, PCplating at 50 Hz in solutions containing appropriate amounts of thiourea and Cl was found to yield macroscopically and microscopicallysmooth deposits with reectance similar to that of a polished uncoated copper substrate. 2004 Elsevier Ltd. All rights reserved.

Keywords: Pulse plating; Copper; Thiourea; Chloride; Morphology

1. Introduction

Typically, dullcopper deposits areproducedbyDC platingin sulphate-plating baths. Two ways to improve their qualityaretheuseofpulseplatingandtheinclusionofadditivesintheplating bath. Pulse current (PC) plating is known to improvethe morphology and properties of deposits in the absenceof additives [15] due to its positive effects on mass trans-

port [68], electrode kinetics [9] and the nucleation of growthcentres [1,10] . Pulse plating also offers a largernumberof pa-rameters (i.e., on-time, off-time, cathodic- and anodic-pulsecurrent density and frequency) than does DC plating to im-prove deposit quality.

Corresponding author. Tel.: +1 519 888 4567x2542;fax: +1 519 746 4979.

E-mail address: [email protected] (M.D. Pritzker).1 Present address: Department of Chemical Technology, Faculty of Sci-

ence, Chulalongkorn University, Bangkok, Thailand.

Deposit properties (i.e., brightness, smoothness and mi-crohardness) can also be improved through the inclusionof additives in a sulphate-plating bath. Both electrocrystal-lization and deposition kinetics are strongly sensitive to thepresence of additives in the plating bath at very low concen-trations. Typically, they inuence deposit morphology andstructure by adsorbing on the cathode and inhibiting variousprocesses during electrodeposition.

Although numerous studies have focused on the inuenceof either pulse plating or additives on copper electrodeposi-tion, relatively few have considered their combined effects[1116] . In most of these reported studies, either the addi-tives were unidentied or interest was focused more on thethickness distribution or throwing power than deposit mor-phology.

Two additives commonly used during copper electrode-position are chloride and thiourea (TU). The presence of small concentrations of chloride as the sole additive in

0013-4686/$ see front matter 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.electacta.2004.08.045


2/13

1850 N. Tantavichet, M.D. Pritzker / Electrochimica Acta 50 (2005) 18491861

sulphate-plating baths has been reported to improve cop-per deposit properties by some researchers [17,18] , but yieldpoorer coatings by others [19,20] . Chloride has also beenincluded in sulphate baths for copper deposition along withother additives [11]. Thiourea is commonly used as a bright-eningagentduring copper deposition [2129] . It functionsby

interacting stronglywith theelectrodesurface andaltering themode of deposition to produce coatings that are microscopi-cally smooth and so appear bright. There is strong evidencethat thiourea forms a bond on the copper surface via its unsat-urated sulfur atom, thereby inhibiting the surface diffusion of the copper adatoms and promoting a smaller-grained struc-ture [26,3034] . In situ atomic force microscopy studies haverevealed that the early stage of electrocrystallization is dom-inated by the formation of at copper islands that grow andeventually merge [35,36] . The edges of these islands appearto act as nuclei for further growth.

The combined effect of thiourea and chloride as additiveshas also been studied. Chloride has been shown to enhancethe adsorption and levelling effect of thiourea during cop-per deposition [37,38] . Investigation by in situ surface en-hanced Raman spectroscopy has revealed that the adsorptionof thiourea onto a copper electrode from a sulphate solution(containing noCu 2+) increasesin thepresenceof Cl [30,32] .

In this study, we investigate the effects of plating modeand the additives chloride and thiourea on the morphologyof copper electrodeposits. The use of PC and pulse reverse(PR) plating over a wide range of frequencies from 50 to50,000 Hz is compared to that of DC plating. Deposit qualityis characterized by scanning electron microscopy (SEM),op-tical microscopy, interference microscopy and scatterometry

and related to cyclic voltammograms for Cu 2+ reduction andelectrode potentials monitored during plating.

2. Experimental

The electrodeposition experiments were conducted usinga rotatingdiscelectrodeimmersed in a cylindrical electrolyticcell containing a 50cm 3 solution.The working electrode wasa 0.635 cm diameter (0.317 cm 2 area) copper disc polishedwith SiC-type abrasive paper (600-grade) and with 0.3 and0.05 m alumina powder to a mirror nish using a motorizedpolisher (Buehler Metaserv Motopol 8). It was then mountedto the end of a teon shaft of a rotating disc assembly (PineInstruments). The counter electrode was a 3.8 cm-diametercopper disc placed at the bottom of the cell located about2 cm from the working electrode.

Pulse plating was generally carried out using a two-electrode system with the working and counter electrodesconnected to a model PARAM4 pulse-plating rectier (LWDScientic). Whenever the working electrode potential wasmonitored, a standard three-electrode system was used andthe electrode response was monitored on a digital oscil-loscope (Agilent 54624A). DC plating and voltammetryexperiments were carried out using an Autolab PGSTAT 10

potentiostat (EcoChemie) and a conventional three-electrodesystem. A mercury/mercurous sulphate electrode (MSE, Ra-diometer Analytical) was used as the reference electrode,although the electrode potentials reported herein were con-verted to the SHE scale.

An acidic sulphate-plating bath consisting of 0.1M

CuSO 4 and 1 MH 2SO4 was used for all experimentsandpre-paredfrom doublydistilled water. To studytheeffectsof addi-tives on copper deposition, various amounts of hydrochloricacid (Aldrich Chemical)or thiourea (Aldrich Chemical) wereused. Stock solutions of thiourea were prepared freshly be-fore the plating experiments. The presence of these additivesin theconcentration rangesused in this work hadonly a negli-gible effect on the open circuit potential (i.e., 0.275 0.01VSHE).

All plating experiments were conducted for 12 min at anaverage current density of 4 A dm 2 and rotational speed of 500 rpm. At this current density and plating time, coatingswith an average thickness of 10.6 m were produced. Thechange in electrode mass over the course of each plating ex-periment was measured. In all cases, this value correspondedto that expected if Cu 2+ reduction to metallic copper was theonly electrode reaction.

Scanning electron microscopy (LEO fuel-emission 1530scanning electron microscope), scatterometry (SMS ScanSystem, Schmitt Industries Inc.), optical microscopy and in-terference microscopy (Veeco, Wyko NT3300) were usedto characterize deposit morphology. Scatterometry (incidentvisible light with a 1300nm wavelength and 25 incidentangle) allowed quantitative measurement of surface bright-ness and RMS microroughness. Polished uncoated copper

discs with a 93% specular reectance and 100200 A RMSroughness were used as a standard to assess the quality of thecopper electrodeposits.

3. Results and discussion

3.1. Polarization curves

Before discussing the effect of additives on depositmorphology, it will be useful to present their effect on theelectroderesponseduring Cu 2+ reduction. Fig.1 shows polar-ization curves for copper deposition obtained in the absenceof additives and in the presence of 274 M HCl (10 ppm),20 M TU and 20 M TU+ 274 M HCl. In the presenceof HCl alone, the electrode potential is depolarized towardmore positive potentials, similar to that reported by others[11,20,34,39] . This enhancement in the reaction rate hasbeen attributed to the formation of a bridge between Cu 2+

and Cl at the electrode surface with a shorter spacing thanthat of Cu2+ H2O metal bridges in chloride-free systems[39,40] .

On the other hand, the voltammograms obtained in thesolutions containing thiourea alone are polarized towardmore negative potential relative to that of an additive-free


3/13

N. Tantavichet, M.D. Pritzker / Electrochimica Acta 50 (2005) 18491861 1851

Fig. 1. Voltammograms for Cu 2+ reduction in: (a) 274 M HCl, (b) 20 MTU and (c) 20 M TU+274 M HCl solutions at 500 rpm and 10 mV s 1

scan rate. Voltammogram for Cu 2+ reduction in solution containing no ad-ditive is also included for comparison.

solution, even at thiourea concentrations as low as 20 M.The voltammogram in Fig. 1 shows almost complete sup-pression of current until the electrode decreases below about 0.05 V SHE. A similar effect has been reported by Farndonet al. [25] and Muresan et al. [29] although for different elec-trolyte compositions. This suppression is presumably due tothe complete coverage by a lm that inhibits Cu 2+ reduction.When the potential becomes more negative than 0.05 V, thelm may be reduced to form copper or break down allowingthe reduction of Cu 2+ from the solution to occur. Once cur-rent begins to ow, it rises very sharply to about 3.5 A dm 2

whereupon it begins to increase more slowly. The limiting

current plateau is much less well dened than in the case of an additive-free solution, as observed previously [25].

When the electrolyte contains both thiourea and Cl ,Cu2+ reduction is suppressed, as in the presence of thioureaalone. However, some difference in the polarization curvesis observed. Once Cu 2+ reduction begins in the presenceof the two additives, the current density rises rapidly allthe way to the limiting current density. A well-dened lim-

iting current density plateau is restored. A second differ-ence arises in the appearance of an additional peak cen-tered at 0.2V. Although not shown here, the height of this peak increases if the amount of Cl added to the elec-trolyte is increased. The process associated with this peak may involve the formation of a thioureaCu(I)chloridelm.

It is worth noting from Fig. 1 that the limiting current den-sity remains essentially the same regardless of the electrolytecomposition. This indicates that Cu 2+ is likely the predomi-nant form of soluble copper involved in copper deposition inall cases. The concentration of any soluble copperthioureacomplex that would form is limited by the amount of thioureain solution and so the limiting current density based on thisform of soluble copper would be far less than that observedin Fig. 1.

3.2. DC plating

TheSEM imagesof thedeposits obtained by DCplating intheabsence andpresenceof additives areshown in Fig.2. The% reectance and microroughness of each deposit obtainedby scatterometry are indicatedon the images. Thedeposit ob-tained by DC plating in the absence of additives appears dull

(38% reectance) with a salmon-red colour surface ( Fig. 2a).In the presence of 274 M HCl alone, the resulting deposit

Fig. 2. SEM images of deposits produced by DC plating at 4 A dm 2 in 0.1M CuSO 41M H2SO4 solutions containing (a) no additive, (b) 274 M HCl, (c)20 M TU and (d) 20 M TU+ 274 M HCl ( 5000 magnication).


4/13


has even poorer quality, as shown in Fig. 2b. It is very dulland non-metallic ( 23% reectance), also with a strongsalmon-red colour and microroughness exceeding 2283 A. Asimilar effect of chloride on deposit roughness has also beenobserved previously via AFM examination [20].

As expected, the addition of 20 M TU to the 0.1M

CuSO 41M H2SO4 solution causes the polarization of theelectrode potential during plating at a current density of 4Adm 2 to increase from 0.063 V SHE to 0.185 V.The SEM image of the deposit obtained under these condi-tions is shown in Fig. 2c. The role of thiourea as a brighteneris clearly evident in that a compact and reective coating(83% reectance) is produced. The coating appearance issimilar to what is expected based on the observations andmechanism proposed by Schmidt et al. [36]. The improve-ment in surface reectance is consistent with their ndingthat thiourea tends to promote two-dimensional growth of at copper islands during electrocrystallization. The smallergrain size is consistent with previous ndings that nucleationduring electrodeposition changes from instantaneous to pro-gressive in the presence of thiourea, with the formation of numerous new nucleation sites [35,41] . However, it shouldbe noted that excessive amounts of thiourea (500 M andabove) are detrimental to deposit morphology by producingswollen, aky and peeling deposits.

Although the deposit is relatively smooth and bright whenthe TU level is 20 M, it exhibits a relatively high micror-oughness (>2283 A) and is not as reective as the polisheduncoated copper substrate due to the appearance of a lightlm. This light lm likely arises due to the formation of mi-croscopic nodules (>1 m diameter, as estimated from SEM

and interference microscopy) over the surface ( Fig. 2c), al-though the deposit elsewhere is smooth and ned-grained.This combination of surface features leads to a bright surfacewitha relatively highmicroroughness. Nodule formationdur-ing copper deposition in the presence of thiourea has beenreported previously [27,36,42] . Schmidt et al. [36] proposedthat once the at copper islands that tend to make the depositvery reective begin to merge, three-dimensional growth oc-curs where these islands intersect. These intersection spotsreceive highercurrent than other areasandultimately developinto microscopic roughness. Although nodules appear in theSEM images, the deposits are still very smooth macroscopi-cally.

Someofthepreviouslyreportedobservationsoftheforma-tion of at copper islands and nodules and the occurrence of electrocrystallization via progressive nucleation were madeduring studies that focused only on the initial stages of de-position on substrates (i.e., glassy carbon and Au) differentfrom that used in the current study. Similarity of the morphol-ogy observed in Fig. 2c to that of the earlier studies indicatesthat these phenomena continue to occur at least up to coat-ing thicknesses of 10 m. Furthermore, the similarities alsosuggest that these phenomena do not depend critically on thenature of the substrate, but more so on interactions betweenthiourea and copper adatoms.

To determine if a combination of the two additives couldhave a synergetic effect on copper deposition, the use of 0.1M CuSO 41M H2SO4 solutions containing 20 M TUand 274 M HCl was studied. The SEM image of the de-posit obtained by DC plating ( Fig. 2d) shows signicant im-provement in deposit microstructure over that produced in the

presence of 20 M TU alone( Fig. 2c) or HCl alone ( Fig. 2b).Nodules areno longer formedanda ne- andsmooth-grainedstructure over the nanometer to micron range is produced.However, the surface becomes so rough on a macroscopicscale, that the % reectance and RMS roughness could notbe accurately measured. No distinct pattern to the roughnessis evident at this HClconcentration. Rough non-uniform cop-perdeposits have also been reported duringDC plating in sul-phate solutions containing chloride and gelatin or bindarine[17].

3.3. Pulsed current (PC) plating

To compare the effectiveness of PC plating to that of DCplating, a series of experiments were conducted to determinethe effect of pulse frequency and duty cycle on coating mor-phology. In order to make a fair comparison, these experi-ments were carried out at a time-averaged current density of 4Adm 2 in solutions with the same compositions as consid-ered in theprevious section. TheSEMimagesof the resultingdeposits are shown in Fig. 3.

In the absence of additives, PC plating leads to a morecompact and smaller-grained structure than does the DCmode, yielding a brighter, smoother and more metallic sur-face (Fig. 3a). These trends agree with the known effects of

pulse plating to enhance metal nucleation [43] and to reducemass transport limitations [1,8,43,44] . Another contributingfactor in improving morphology is the likely effect of plat-ing mode on the deposition mechanism. Electrocrystalliza-tion during DC plating of copper in the absence of additivestends to occur by an instantaneous nucleation mechanism[35,36,41] . With the step change in electrode potential in thenegative direction at the beginning of each pulse cycle, PCplating likelyshifts the mechanismtoward that of progressivenucleation where nuclei form continually throughout depo-sition leading to a ner and smaller-grained structure.

Two other noteworthy trends in Fig. 3a are the effectsof pulse frequency and duty cycle during PC plating. For agiven duty cycle, % reectance and deposit microsmooth-ness improve as the pulse frequency is increased from 50 to500 Hz, but then begin to deteriorate at higher frequencies.This trend likely reects the competing effects of pulse dura-tion and amplitude. With an increase in frequency from 50 to500Hz, the larger number of pulsesoutweighs the simultane-ous diminution of their amplitudes and produces a smaller-grained and brighter surface. This idea is supported by theresults in Fig. 4a and b that show the measured electrode po-tentialsduring the on-time are similar at50and 500 Hz. How-ever, as the frequency is increased above 500 Hz, the smalleramplitude of the electrode potential pulse due to double layer


5/13


Fig. 3. SEM images of deposits produced by PC plating at different frequencies and duty cycles in solutions of 0.1 M CuSO 41M H2SO4 containing (a) noadditive, (b) 20 M TU and (c) 20 M TU+ 274 M HCl ( 5000 magnication).

charging begins to outweigh the effect of shorter pulse du-rations (Fig. 4c). Although the electrode response becomesmore DC-like as frequency is raised, it is still not completelydistorted to a DC signal at a frequency as high as 50 kHz andthe corresponding deposit is still brighter and smoother thanthat obtained by DC plating.

Comparison of the coatings in Fig. 3a reveals that lowerduty cycles yield brighter and smoother deposits at 50 Hzfrequency. This is presumably linked to the current and elec-trode potential reached duringeach cathodic pulse.Examina-tion of the corresponding electrode responses (not included

here) gives the expected result that the potentials reach morenegative values at lower duty cycle, something that shouldproduce a ner-grained structure. This effect should be morepronounced at lower frequencies than at higher frequenciesand is reected in the corresponding deposit morphologies(not included here).

Whereas the plating mode has a signicant effect whenplating is conducted in the absence of additives, it has littleeffect when 274 M HCl alone is added to the solution. Poorquality deposits very similar in nature to that shown in Fig.2bfor DC conditions were obtained at all duty cycles (20, 50


6/13


Fig. 4. Comparison of electrode potentials monitored during PC plating in 0.1 M CuSO 41M H2SO4 solutions with and without additives at 4 A dm 2 average

current density, frequencies of 50, 500 and 50,000 Hz and duty cycles of (ac) 20% and (df) 50%.

and 80%) and pulse frequencies (50, 500 and 50,000Hz)investigated.

SEM images of the deposits obtained by PC plating insolutions with 20 M TU alone at various frequencies andduty cycles are shown in Fig. 3b.The deposit structure differsdepending onthe dutycyclefor PCplating at50 Hz. At50 and80% duty cycles, deposit appearance improves signicantly(up to 8085% reectance) compared to that produced byPC plating in additive-free solutions. These deposits containverysmall-grained crystals, although nodules observed in thecoatings produced under DC conditions are still observed.

On the other hand, the deposit produced in a 20 M TUsolution at 20% duty cycle has low reectance (58%) with arough and coarse-grained structure. Moreover, it has a non-

uniform thicknesswith a relatively bright region near the disccentre that becomes thicker and duller near the edge. Thisdeposit is poorer in quality than that produced in an additive-free solution under the same plating conditions and in 20 MTU solutions at higher duty cycles. This trend is opposite tothat observed in the additive-free system where PC plating atthe lowest duty cycle produces the brightest and smoothestdeposit. The poor deposit could be caused by mass trans-fer limitations due to the application of a very high cathodicpulse current density at the low duty cycle. For PC plating at20% duty cycle and 50 Hz, the current density of 20 A dm 2

during the cathodic portion of the pulse cycle is close to thelimiting cathodic pulse current density of 22.6 A dm 2 calcu-lated from the expression presented by Ibl [9] and Chene and


7/13


Landolt [8]. With thiourea present, the electrode potential issubstantially polarized to about 0.4V SHE, as shown inFig. 4a. Comparison with the corresponding voltammogramin Fig.1 b shows that thesystemis close to mass transfer limit-ing conditions at this potential. Consequently, a poor depositwould be expected. This explanation is supported by a PC

plating experimentcarried outat a rotational speedof 750rpmunder otherwise identical conditions that yields an improve-ment in deposit quality (80% reectance and>2283 A micro-roughness), similar to that obtained at duty cycles of 50 and80% at 50 Hz. It is not obvious why deposit quality shouldbe more prone to mass transfer limitations at 20% duty cyclewhen thiourea is present than when it is absent. It is possiblethat the desorption of the copperthiourea complex from theelectrode during electrodeposition which may be critical tothe resulting morphology, becomes mass transfer controlledat this duty cycle.

However, when frequency is raised to 500 Hz and above,duty cycle and frequency no longer affect morphology sig-nicantly ( Fig. 3b). All deposits are smooth, compact and re-ective without the formation of a rim along the disc edges. Alight lm appears as in DC plating, reducing the reectanceslightly from that of the polished uncoated substrate. Thesedeposits appear to be similar to those produced by DC plating(Fig. 2c) and PC plating at 50 Hz (except at 20% duty cycle)at the same thiourea level.

Comparison of these trends with those of the correspond-ing electrode responses in Fig. 4 shows some interestingeffects. Similar deposit morphologies are obtained in thepresence of 20 M TU during DC- and PC plating at 50 Hz(except at 20% duty cycle) although the electrode responses

are vastly different. At higher frequencies of 500Hz and50 kHz, polarization during the on-time in the presence of thiourea is similar to that in an additive-free solution, unlikethe case at 50 Hz. Despite the similar electrode responsesobtained in these solutions, the resulting morphologies arenot at all alike. Overall, the additive has a larger effect onthe resulting morphology than plating mode and electroderesponse in these cases.

As discussed previously, the addition of 274 M HCl toa solution containing 20 M TU prevented the formation of nodules that had been observed in the coatings produced byDC plating when thiourea alone was present. Unfortunately,this led to a considerable amount of macroroughness. PCplating experiments were therefore conducted to determineif this macroroughness could be eliminated. The SEM im-ages of the deposits so obtained are shown in Fig. 3c. As inthe case of DC plating ( Fig. 2d), a ne- and smooth-grainedstructure is produced with no evidence of microscopic nod-ules regardless of the pulse frequencies and duty cycles used.Although the deposits are microscopicallybright and smooth,their roughness on a macroscopic scale depends strongly onfrequency.

The optical images of the deposits obtained at 50% dutycycle and frequencies of 50, 500 and 50,000 Hz in a solu-tion containing 20 M TU and 274 M HCl are presented in

Fig.5a. Also included forcomparison is thecoating producedunder DCconditionsat thesameelectrolyte composition. Thesurface obtained at 50Hz exhibits a spiral pattern and differsfrom the one obtained by DC plating that is rough with nodistinct pattern. However, when the frequency is increased,spirals are no longer observed, but other rough patterns are

formed.An effort to eliminate the macroscopic roughness wasmade by varying the amount of HCl in solutions containing0.1M CuSO 4, 1M H2SO4 and 20 M TU. At HCl concen-trations of 27.4 (not shown) and 1 103 M, the depositsproduced by DC plating remain macroscopically rough, butin different ways. Pits are distributed over the surface at thelower HCl concentration, while deep ridges spiral out fromthe centre at the higher HCl level ( Fig. 5b).

Next, PC plating was carried out in solutions containing0.1M CuSO 4, 1M H2SO4 , 20 M TU and higher HCl con-centrations. The optical images of the deposits obtained insolutions containing 1 103 and 2.74 103 M HCl at 50%duty cycle are shown in Fig. 5b and c, respectively. Roughdeposits are produced at frequencies above 50 Hz, while verysmoothdeposits areobtainedat 50 Hz.Althoughnot includedhere, similarly smooth deposits to that obtained at 50% dutycycle are produced by 50 Hz PC plating at 20 and 80% dutycycles.

The deposits obtained at 1 103 M HCl appear betterthan thoseobtainedunder thesame plating conditions (50Hz)at 2.74 103 M HCl. The surfaces obtained at 1 103 MHCl are very bright and reective without the trace of thelight lm observed when thiourea alone is added or the whitelm when 20 M TU+2.74 103 M HCl are added. This

surface exhibits the best nish produced in this study, with areectance and RMS roughness of 92% and 250 A, respec-tively. This is very close to the nish of the polished uncoatedcopper substrate (9394% reectance and 100200 A RMSroughness). An SEM image of this sample was obtained at amuch higher magnication of 50,000 ( Fig. 6a). The depositstill appears to be ne grained, uniform and smooth. FromFig. 6a, the grain size of the coating obtained at 1 103 MHCl is estimated to be on the order of 50 nm in diameter. Atthehigher HClconcentrationof 2.74 103 M,thegrainsizeis coarser and the coating exhibits a light white lm, giving ita slightly lower reectance (87%) and a substantially higherRMS roughness (1300 A) (Fig. 5c). This is conrmed in thecorresponding high magnication SEM image in Fig. 6b.

These results provide further support that chloride playsa major role during electrocrystallization since an excess ortoo little amount tends to coarsen the structure. Yoon et al.[24] observed a similar effect during DC plating with brightdeposits obtained at 1.4 103 M Cl , but dull depositsat lower and higher Cl concentrations. The deleterious ef-fect of chloride at concentrations above 2.8 103 M hasbeen observed by other researchers in systems where it ispresent alone [45,46] or with other additives [17]. Yoon etal. [24] also obtained evidence that the effect of Cl concen-tration on deposit morphology may be related to changes in


8/13


Fig. 5. Optical microscope images of deposit macrostructures produced by DC and PC plating under various conditions in solutions containing 0.1 M CuSO 4 ,1 M H2SO4 , 20 M TU and (a) 274 M HCl, (b) 1 103 M HCl and (c) 2.74 103 M HCl at a rotational speed of 500rpm.

copperchloride chemistry. They showed that when the Cl

concentration is raised above about 845 M, the Cu(I) Clcomplex produced changes from a soluble form to an insol-uble form on the cathode surface. Thus, the light white lmformedon thedeposit in a solution containing 2.74 103 MHCl (Fig. 5c) may be insoluble CuCl.

Fig.6. Highmagnication SEMimagesof depositmicrostructures producedby PC plating at 50 Hz and 50% duty cycle in solutions containing 0.1MCuSO 4 , 1M H2SO4 , 20 M TU and (a) 1 103 M HCl and (b) 2.74 103 M HCl at a rotational speed of 500rpm ( 50,000 magnication).

An important observation from the current study is thatadditive concentration alone does notdetermine deposit mor-phology in this case since a coating as smooth and reectiveas the polished uncoated substrate could only be obtainedthrough use of PC electrolysis. This contrasts with the be-haviour noted earlier whereby the effect of additive concen-tration dominatesover that of theplating mode when thioureaalone is present.

3.4. Spiral formation

In order to more closely characterize the extent and geom-etry of the spiral patterns formed in the copper coatings, thedeposits were examined using interference microscopy. Anexample of the typical geometry of these spirals is shown inFig. 7. These images were obtained for the sample originallypresented in Fig. 5a produced by PC plating at 50 % duty cy-cle and 50 Hz in a solution containing 20 M TU+274 MHCl. Fig. 7a shows the top view, while Fig. 7b shows a three-dimensional perspective of this view. Variations in verticalheight of the surface along a 4.6 mm line in one direction and3.5 mm line in another are presented in Fig. 7c. The spiral-patterned ridges vary from 4 to 8 m in height, which is sig-nicant giventhe average thicknessof thecoating is10 m.In


9/13


Fig. 7. Interference microscope images of deposit produced by PC plating at 50 Hz and 50% duty cycle in a solution containing 20 m TU+274 M HCl:(a) top view ( 1.25), (b) 3D perspective ( 1.25), (c) height proles across the length and width of sample ( 1.25), (d) 3D perspective ( 100) and (e) heightproles ( 100).

contrast, the imagesobtained at higher magnication ( 100)(Fig. 7d and e) reveal a microscopically smooth deposit withundulations reecting the macroscopic roughness.

Thiourea has been shown to act much more as a brighten-ing agent than as a levelling agent in the absence of Cl [27].It is clear from the results in Figs. 5 and7 that the presence of Cl tends to heighten this difference regardless of electrolytecomposition and plating mode.

As shown in this study, coatings with a spiral pattern canform under a range of conditions, but require the presenceof both thiourea and chloride. Spiral-pattern coatings havealso been obtained when a combination of PEG and chloride

[47,48] and gelatin alone [49] have been used during depo-sition of copper and other metals such as nickel [50,51] , zinc[49,52,53] and silver [49], although the mechanism is notwell understood. In our case, it is not likely caused by thetraces of wakes due to hydrogen bubbles being ung off therotating disc, as in other systems [50,51] , since the presenceof thiourea and chloride tends to inhibit hydrogen evolution[54] and no hydrogen evolution was evident during our plat-ing experiments. Spiral formation is also not caused by masstransport limitations since it still occurs when the rotationalspeed is increased from 500 to 2000 rpm. Instead, it is likelyrelated to a change in electrode kinetics due to the interaction


10/13


of chloride, copper and thiourea and the ow patterns nearthe rotating disc electrode.

The most relevant explanation of spiral formation to oursystem was proposed by Hill et al. [48] and Hill and Rogers[47] who also studied copper deposition in acidic sulphatesolutions. As in our case, spirals formed only when chloride

was present and the polarization curves showed a completesuppression of metal reduction up to some critical potentialfollowed by an abrupt rise in current density to the valuesobtained in additive-free solutions (as in Fig. 1c). Accordingto Hill and Rogers [47], a complex thioureaCu(I)chloridelm can exist on the electrode surface below the critical po-tential and prevent copper deposition. However, when thispotential is reached, surface coverage by this lm decreasesrapidly allowing the current to rise abruptly. Thenature of theresulting deposit pattern is closely related to the position onthe steep current rise portion of the polarization curve wherethesystemoperates. In our study, we conductedDC plating at4Adm 2 near the upper end of the steep current rise. At thestart of deposition, small protrusions and depressions only afew molecules thick will inevitably form on theelectrodesur-face. The local current density at protrusions will be greaterthan at the surroundings, whereas the current density at de-pressions will be lower. When the system operates near theupper end of the current rise, the lm breaksdownand coppertends to deposit everywhere except in the depressions. These

depressions are reinforced by the rotating uid ow, enablingmacroscopic spirals in the deposit to develop. However, if thesystem operates above the current rise and the critical poten-tial, the lm breaks down completely and spiral patterns donot form.

The situation during PC plating is complicated by the fact

that the adsorption/desorption behaviour of thiourea both inthe presence and absence of Cl depends on electrode poten-tial [3033] . Consequently, this factor may play an importantrole during each pulse cycle as the applied current and result-ing electrode potential change. Depending on the operatingconditions and the pulse parameters, the properties and theresponses of a complex thioureaCu(I)chloride lm to theapplied current could differ during DC and PC plating, lead-ing to different deposit topographies.

We assessed the reproducibility of spiral formation by re-peating the various experiments where they were rst ob-served to occur. Spirals always form, although they vary intheir extent over the surface and the depth of their ridges.In some cases, the spiral pattern was conned to the cen-tral region with an unpatterned but rough deposit toward theedge. The current at the edge is higher than at the centre andmay be high enough to exceed the critical conditions. As aresult, it would be more difcult for spirals to form near theedge. The deposit structure produced by PC plating in solu-tions containing 20 M TU and 274 M HCl changes from

Fig. 8. SEM images of deposit microstructures produced by PR plating at 4 A dm 2 average current density, cathodic and anodic pulse current densities of 6 and 18A dm 2 , respectively, at different frequencies in solutions containing 0.1 M CuSO 4 , 1M H2SO4 and (a) no additive, (b) 20 M TU and (c) 20 MTU+274 M HCl ( 5000 magnication).


11/13


a spiral pattern to an unpatterned-rough surface as frequencyincreases. Again, this is consistent with theeffect of electrodepotential proposed above since the potentials reached duringthe on-times at higher frequencies are less negative than at50 Hz. At the highest pulse frequency studied (50 kHz), thedeposit appears similar to that obtained by DC plating due to

the double layer effect.

3.5. Pulse reverse (PR) plating

We conducted PR plating at a time-averaged current den-sity of 4Adm 2 (same as during the DC and PC plating

experiments) and cathodic and anodic pulse current densitiesof 6 and 18 A dm 2, respectively (i.e., duty cycle of 91.67%)in solutions containing 20 M TU with and without HCl.Experiments conducted at an anodic pulse current densityof 6Adm 2 (i.e., cathodic:anodic pulse current ratio of 1:1and duty cycle of 83.33%) yielded similar results for most

conditions and so are not included here.SEM images of the deposits obtained by PR platingat the three frequencies in the different plating solutionsare shown in Fig. 8. Much rougher and coarser depositsare obtained at 50Hz than at the higher frequencies ineach solution. The poor quality obtained at 50 Hz is likely

Fig. 9. Comparison of electrode potentials monitored during PR plating at 4A dm 2 average current density, cathodic and anodic pulse current densities of 6and 18A dm 2 , respectively, at different frequencies in solutions containing 0.1 M CuSO 4 , 1M H2SO4 and (ac) 20 M TU and (df) 20 M TU+274 MHCl. The electrode potential for Cu 2+ reduction in a solution containing no additive is included in each gure for comparison.


12/13


related to the electrode responses during PR plating. Ata frequency of 50 Hz, the electrode potential rises abovethe open-circuit potential during the anodic pulse indicat-ing that some anodic dissolution of copper occurs ( Fig. 9).During the reverse-time, small grains may dissolve prefer-entially since they are less thermodynamically stable than

larger grains [4]. On the whole, this would make the coat-ing more coarse-grained. Unlike what is observed during DCand PC plating at this frequency, the presence of thioureaworsens deposit quality from that in an additive-free solu-tion. The coating structure in the presence of 20 M TUwith and without HCl is very coarse and cauliower-like.Macroscopically, it has a non-metallic appearance, with astrongsalmon-red or almostbrick-redcolour andvirtuallynoreectance.

An increase in frequency to 500 Hz during PR plating im-provesdeposit propertieswhen no additive ( Fig.8 a)or20 MTU alone (Fig. 8b) is present. In fact, the deposits obtainedin the presence of thiourea alone are now very similar in ap-pearance to those obtained by DC and PC plating. On theother hand, a deposit similar to that observed at 50 Hz is ob-tained for a plating solution containing 20 M TU + 274 MHCl (Fig. 8c). Examination of the corresponding electroderesponses in Fig. 9b and e helps explain this effect. The elec-trode potentialduring theanodicpulse is suppressed at higherfrequencies and does not reach as positive a value as at 50 Hzdue to the double layer effect. The electrode potential duringthe anodic pulse in a solution containing 20 M TU does notreach the open-circuit potential suggesting that no metal dis-solution occurs, as in the case of PC plating. Not surprisingly,the resulting deposit morphology is similar to that obtained

by PC plating at the same frequency and is considerably bet-ter than that achieved by PR plating at 50 Hz ( Fig. 8a). Incontrast, some dissolution still occurs during PR plating at500Hz in an additive-free solution ( Fig. 9b) and a solutioncontaining 20 M TU+274 M HCl (Fig. 9e). Some im-provement in deposit quality is observed in the additive-freesolution by an increase of frequency to 500Hz ( Fig. 8a),although not as much as by DC or PC plating. However,in the case of PR plating in a solution containing 20 MTU+274 M HCl, no substantial improvement of morphol-ogy is observed from the deposit produced by PR plating at50 Hz. The correlation between the potentials reached duringthe anodic pulse and deposit morphology indicates that cop-per dissolution is deleterious to deposit quality, at least forthese experiments.

When pulse frequency is further increased to 50 kHz, theelectrode potential during the anodic pulse is suppressedmore and, in fact, never reaches the open-circuit potentialin all plating solutions. The electrode response becomes sim-ilar to that obtained during PC plating at the same frequency.Comparison of Fig.8 with Fig.3 shows that theresulting mor-phologies are similar to those obtained during PC plating inthe respective solutions. This further supports the conclusionthat dissolution of copper during PR plating is harmful todeposit quality.

4. Conclusions

PC plating in additive-free solution can improve copperdeposit properties over those achieved by DC plating. Of the conditions investigated, PC plating at 500 Hz and 20%duty cycle was found to produce the best deposits. How-

ever, with thiourea present, the deposits obtained by DC andPC plating were similar, except for PC plating at 50 Hz and20% duty cycle. This indicates that when thiourea is the onlyadditive, its presence has a stronger effect than the platingmode. Overall, it leads to a higher quality deposit than thatobtained with no additive, but still not as good as the pol-ished uncoated substrate. The deposit has a thin cloudy lmwith high microroughness due to microscopic nodules dis-tributed over its surface. DC plating in solutions containingthiourea and chloride tends to produce copper deposits withbrightness close to that of the polished undeposited substrate,but with signicant macroroughness. By applying PC plat-ing under suitable conditions (50Hz in solutions containing20 M TU+ 1 103 M HCl), smooth and featureless de-posits with brightness close to that of the polished substrateareproduced. Poorerdepositsareobtainedby PR plating bothin the presence and absence of additives than those obtainedby PC and even DC plating whenever the electrode potentialrises above the open-circuit value during the anodic portionof the pulse cycle and dissolution of copper presumably oc-curs. At higher frequencies, the double layer effect becomesimportant and the electrode response and deposit morphol-ogy become similar to that observed during high frequencyPC or DC plating.

Acknowledgement

Theauthors express their gratitude to theNatural Sciencesand Engineering Research Council of Canada (NSERC) fornancial support during the course of this study.

References

[1] J.C. Puippe, N. Ibl, Plat. Surf. Finish 6 (1980) 68.[2] W. Kim, R. Weil, Surf. Coat. Technol. 38 (1989) 289.[3] A.M. El-Sherik, U. Erb, J. Page, Surf. Coat. Technol. 88 (1996) 70.

[4] S. Roy, D. Landolt, J. Appl. Electrochem. 27 (1997) 299.[5] P. Kristof, M. Pritzker, Plat. Surf. Finish 75 (1998) 237.[6] D.-T. Chin, J. Electrochem. Soc. 130 (1983) 1657.[7] M. Datta, D. Landolt, Surf. Technol. 25 (1985) 97.[8] O. Chene, D. Landolt, J. Appl. Electrochem. 19 (1989) 188.[9] N. Ibl, Surf. Technol. 10 (1980) 81.

[10] H.Y. Cheh, J. Electrochem. Soc. 118 (1971) 551.[11] T. Pearson, J.K. Dennis, J. Appl. Electrochem. 20 (1990) 196.[12] T. Pearson, J.K. Dennis, Surf. Coat. Technol. 42 (1990) 69.[13] M.S. Aroyo, Plat. Surf. Finish 82 (1995) 53.[14] D.S. Stoychev, M.S. Aroyo, Plat. Surf. Finish 84 (1997) 26.[15] S.-C. Kou, A. Hung, Plat. Surf. Finish 87 (2000) 140.[16] S.S. Kruglikov, P. Becker, O. Jankowski, Plat. Surf. Finish 89 (2002)

45.[17] W.H. Gauvin, C.A. Winkler, J. Electrochem. Soc. 99 (1952) 71.


13/13


[18] A. Radisic, A.C. West, P.C. Searson, J. Electrochem. Soc. 149 (2002)C94.

[19] E.W. Rouse, P.K. Aubel, Trans. Electrochem. Soc. 52 (1927) 189.[20] M. Kang, A.A. Gewirth, J. Electrochem. Soc. 150 (2003) C426.[21] B. Ke, J.J. Hoekstra, B.C. Sison, D. Trivich, J. Electrochem. Soc.

109 (1962) 798.[22] A. Szymaszek, J. Biernat, L. Pajdowski, Electrochim. Acta 22 (1977)

359.[23] P. Cofr e, A. Bustos, J. Appl. Electrochem. 24 (1994) 564.[24] S. Yoon, M. Schwartz, K. Nobe, Plat. Surf. Finish 81 (1994) 65.[25] E.E. Farndon, F.C. Walsh, S.A. Campbell, J. Appl. Electrochem. 25

(1995) 574.[26] S.A. Campbell, E.E. Farndon, F.C. Walsh, M. Kalaji, Trans. Inst.

Met. Fin. 75 (1997) 10.[27] M.A. Alodan, W.H. Smyrl, J. Electrochem. Soc. 145 (1998) 957.[28] A. Tarallo, L. Heerman, J. Appl. Electrochem. 29 (1999) 585.[29] L. Muresan, S. Varvara, G. Maurin, S. Dorneanu, Hydrometallurgy

54 (2000) 161.[30] B.H. Loo, Chem. Phys. Lett. 89 (1982) 346.[31] G.M. Brown, G.A. Hope, D.P. Schweinberg, P.M. Fredericks, J. Elec-

troanal. Chem. 380 (1995) 161.[32] G.M. Brown, G.A. Hope, J. Electroanal. Chem. 413 (1996) 153.

[33] D. Papapanayiotou, R.N. Nuzzo, R.C. Alkire, J. Electrochem. Soc.145 (1998) 3366.[34] J.J. Kelly, A.C. West, J. Electrochem. Soc. 145 (1998) 3472.[35] M.H. H olzle, C.W. Apsel, T. Will, D.M. Kolb, J. Electrochem. Soc.

142 (1995) 3741.

[36] W.U. Schmidt, R.C. Alkire, A.A. Gewirth, J. Electrochem. Soc. 143(1996) 3122.

[37] A. Szymaszek, L. Pajdowski, J. Biernat, Electrochim. Acta 25 (1980)985.

[38] J.D. Reid, A.P. David, J. Electrochem. Soc. 134 (1987) 1389.[39] Z. Nagy, J.P. Blaudeau, N.C. Huang, L.A. Curtis, D.J. Zurawski, J.

Electrochem. Soc. 142 (1995) L87.

[40] S. Goldbach, W. Messing, T. Daenen, F. Lapicque, Electrochim. Acta44 (1998) 323.[41] G. Fabricius, K. Konttiri, G. Sundholm, Electrochim. Acta 39 (1994)

2353.[42] D.F. Suarez, F.A. Olson, J. Appl. Electrochem. 22 (1992) 1002.[43] N. Ibl, J.C. Puippe, H. Angerer, Surf. Technol. 6 (1978) 287.[44] D. Landolt, in: J.C. Puippe, F.H. Leaman (Eds.), Theory and Practice

of Pulse Plating, AESF, Orlando, Fla., 1986 (Chapter 5).[45] Y.-L. Yao, Trans. Electrochem. Soc. 86 (1944) 371.[46] R. Walker, S.D. Cook, Surf. Technol. 11 (1980) 189.[47] M.R.H. Hill, G.T. Rogers, J. Electroanal. Chem. 86 (1978) 179.[48] M.R.H. Hill, G.T. Rogers, K.J. Taylor, J. Electroanal. Chem. 96

(1979) 87.[49] J. Jorn e, M.G. Lee, J. Electrochem. Soc. 143 (1996) 865.[50] G.T. Rogers, K.J. Taylor, Nature 200 (1963) 1062.

[51] G.T. Rogers, K.J. Taylor, Electrochim. Acta 13 (1968) 109.[52] C. Cachet, Z. Chami, R. Wiart, Electrochim. Acta 32 (1987) 465.[53] K.E. Yee, J. Jorne, J. Electrochem. Soc. 137 (1990) 2403.[54] J. Bukowska, K. Jachowska, J. Electroanal. Chem. 367 (1994)

41.

20 effect of plating mode, thiourea and chloride on the morphology of.pdf

Documents