Electrochimica Acta 47 (2001) 61–66

www.elsevier.com/locate/electacta

Pulse reversal plating of nickel and nickel alloys formicrogalvanics

Peter T. TangDepartment of Manufacturing Engineering (IPT), Technical Uni�ersity of Denmark, Building 204, 2800 Lyngby, Denmark

Received 1 October 2000; received in revised form 20 March 2001

Abstract

The use of pulse reversal (PR) plating, as an alternative to the use of additives for electrochemical deposition of nickel, isstudied. With optimised pulse plating parameters, and in some cases in combination with additives, substantial improvement ofthe deposit properties can be achieved. Utilising chloride-type baths, PR plating of pure nickel (and harder nickel–cobalt alloys)have been used to fabricate tools for micro-injection moulding (Polymer Structures for �Tas, EuroSensors, Copenhagen, 27–30August, 2000) and micromechanical structures. Furthermore, preliminary results from pulse plating experiments with ternarymagnetic alloys, comprising 50–60% Co, 25–35% Fe and 10–20% Ni, will be reported. For both pure nickel and nickel alloysgood chemically stable electrolytes have been developed, and the deposits are smooth with low residual stress. None of theelectrolytes contain sulphur co-depositing additives (such as saccharin) nor wetting agents. © 2001 Elsevier Science Ltd. All rightsreserved.

Keywords: Microgalvanics; Nickel–cobalt; Cobalt–nickel–iron; Pulse plating

1. Introduction

Pure nickel, deposited from a sulphamate bath, hasuntil recently been the preferred material for microgal-vanics, i.e. microelectromechanical systems (MEMS)and for electroforming of parts with micrometerdimensions.

The sulphamate nickel process, however, is not with-out problems. In particular, it can be difficult to avoidpitting (defects in the deposits created by hydrogenbubbles) and at the same time obtain smooth surfaces,good distribution of material and low stress over largerareas.

Pulse plating has previously been reported to im-prove the properties of nickel and nickel alloy deposits.Typically, focus has been on properties such as grainsize [2] and smoothness [3]. When pulse plating is to beutilised for microgalvanics, however, internal stress andmaterial distribution (both on micro and macro level) iseven more important.

To be able to switch rapidly between deposition anddissolution of nickel (pulse reversal (PR) plating), with-out risking passivation, a bath consisting of nickelchloride, nickel sulphate and boric acid (Watts-typeelectrolyte but with more chloride than sulphate) isused [4].

The same type of electrolyte has also been used fordeposition of NiCo and CoNiFe alloys. In these cases,the chloride salts of the alloying elements are used,typically replacing a similar amount of nickel chloride.The main motivation for the development of PR plat-ing processes for NiCo and CoNiFe is that this methodprovides deposit with low residual stress without co-de-position of sulphur [5]. Even small amounts of sulphur(co-deposited with NiCo) reduce corrosion resistance.In addition, since sulphur will diffuse to the grainboundaries if the alloy is heated (thereby making theNiCo alloy hard and brittle), the material becomesuseless for micro-injection moulding tools [1]. Similarly,co-deposition of sulphur in CoNiFe alloys will alsoreduce corrosion resistance and change (usually in anegative way) the magnetic properties of this soft-mag-netic material [6].E-mail address: tt@ipt.dtu.dk (P.T. Tang).

0013-4686/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.PII: S 0 0 1 3 -4686 (01 )00578 -3

P.T. Tang / Electrochimica Acta 47 (2001) 61–6662

Fig. 1. Typical PR waveform, with the four pulse plating parametersneeded to define one cycle (and indication of the anodic and cathodiccharge involved).

Fig. 2. Cross section of the corner of two steel panels plated with purenickel (150 �m average thickness) using either DC or PR plating. Theburr at the panel corner is created by the cutting machine. It isobvious that the PR plated deposit has a better material distribution,and that it appears to be without pores or other mechanical weak-nesses.

Any electroplated soft magnetic alloy is usually com-pared to PermAlloy (NiFe, 21 at.% Fe) which untilrecently was the only electroplated magnetic materialfor MEMS applications. In order to reduce the size ofMEMS devices, or to enhance the properties of existingdevices, it is necessary to increase the saturation fluxdensity.

Osaka et al. [6] have reported electroplated CoNiFefilms with saturation flux densities (Bs) up to 2.1 T —and with a relatively low coercivity (Hc) of less than 2.0Oe (�160 A/m). This is very good compared to thebest recorded values for PermAlloy (Bs=1.1 T andHc=10 A/m) [7].

According to Osaka the highest Bs-values are ob-tained when the alloy is fine grained — especiallyaround the bcc to fcc phase boundary. The exactposition of this phase boundary (see Fig. 6) depends onthe additives used in the electrolyte. When saccharin isused to control the stress of the deposits, sulphur willco-deposit with the alloy creating a CoNiFeS materialwith somewhat different properties. This co-depositionof sulphur will also move the phase boundary [6].

2. Experimental

Two types of experimental set-up have been used.For initial experiments with bath composition, hardnessmeasurements etc., a small bath volume of 2500 ml in aglass container was used. A magnetic stirrer providedagitation (and heating) and the substrate was a 30×30mm copper foil. The anode was pure nickel (even forthe alloys) mounted on a titanium cord. Since theelectrolytes were never used for very long, it is reason-able to assume that the bath content was not changedsignificantly.

In order to measure residual stress (by the wafercurvature method [7]), 4 in. silicon wafers with a 200/1000 A� Ti/Au plating base were used. For these largersubstrates 25 l electrolytes were used, fitted with aceramic heater, continuous filtration and a titaniumbasket as the anode. For pure nickel and NiCo bathscompressed air agitation was utilised for agitation andfor the CoNiFe process, mechanical movement of bothanode and cathode was used. The anode basket con-tained 100% pure nickel pieces for the pure nickel bathand a mixture 50% nickel and 50% cobalt for the alloys.

Table 1Chemical composition of the electrolytes used

0.186 0.170Nickel sulphate 0.100Nickel acetate 0.170Nickel chloride 1.2501.262 0.310Cobalt chloride 0.068 0.240Iron chloride 0.160

0.650Boric acid 0.650 0.6500.010Naphthalene 1,3, 6-trisulphonic acid 0.010 0.010

0.0405-Sulphosalicylic acid

NiCo electrolytes with either 0.170 M nickel sulphate or 0.170 Mnickel acetate, and with 0.034 or 0.068 M of cobalt chloride (corre-sponding to 2.0 or 4.0 g/l of cobalt) were studied. Several versions ofthe CoNiFe bath were examined; the composition listed correspondsto series IV in Fig. 6.

P.T. Tang / Electrochimica Acta 47 (2001) 61–66 63

Fig. 3. SEM micrograph of DC plated NiCo alloys plated from three different electrolytes (sulphate bath with 2.0 g/l of Co, sulphate bath with4.0 g/l of Co and acetate bath with 4.0 g/l of Co) and at a current density of 1.0 or 2.5 A/dm2.

Fig. 4. SEM micrograph of PR plated NiCo alloys plated from three different electrolytes (sulphate bath with 2.0 g/l of Co, sulphate bath with4.0 g/l of Co and acetate bath with 4.0 g/l of Co) and at an average current density of 1.0 or 2.5 A/dm2.

All electrolytes were operated at 40 °C and at a pHvalue of 4.0. In all PR experiments the cathode time (tc)was 60 ms and the anode time (ta) 20 ms. The cathode(ic) and anode (ia) current density was varied, but the(iata)/(ictc) ratio was maintained at 50% with ia=1.5ic(see Fig. 1). The average current density (iav) is theniav= (ictc− iata)/(tc+ ta).

The composition of the various electrolytes can befound in Table 1. For NiCo two versions, with nickelsulphate or with nickel acetate, have been tested withcobalt concentrations of 2.0 g/l (0.034 M) and 4.0 g/l(0.068 M).

The CoNiFe alloy electrolyte was changed stepwisetowards the one listed in Table 1 (called series IV), sothat the metal concentrations for series I (see Fig. 6)was 0.200 M Co, 0.080 M Fe and 0.500 M Ni.

Alloy compositions were determined by EDX mea-surements and hardness by the micro Vickers methodwith a 100-g load.

Fig. 5. Cobalt in the deposit as a function of average current densityfor NiCo alloys plated from different electrolytes (see Table 1).

P.T. Tang / Electrochimica Acta 47 (2001) 61–6664

Fig. 6. A section of the CoNiFe system with indication of thecomposition of the various alloys deposited. The phase boundarybetween bcc (body centred cubic) and fcc (face centred cubic) dependson the use of additives [6]. The most fine-grained materials, with thehighest saturation flux density are expected to be found near thisphase boundary (adapted from Osaka et al. [6]).

3.1. NiCo

For the NiCo alloys a number of SEM micrographscaptured at the same magnification have been com-pared. Looking at direct current (DC) plating for threedifferent electrolytes at two current densities (Fig. 3) itappears that the current density (in the range investi-gated) has only an insignificant influence on the surfacetopography. Higher amounts of cobalt in the sulphatebath seem to improve the appearance of the deposits,and this corresponds well with the fact that historicallycobalt has been used as a brightener for nickel baths.The acetate bath produces deposits that are generallymore rough than deposits from the sulphateelectrolytes.

Introducing PR plating for the same electrolytes, andwith the same average current density levels (Fig. 4),the picture changes entirely. The deposits from thesulphate baths seem to be electropolished and have asemi-bright appearance. The bath with 4 g/l of cobaltstill provides the smoothest deposits of the sulphatebaths. The acetate bath, with the PR plating, com-pletely changes to a mirror bright surface far smootherthan any other NiCo deposit. The explanation lies inthe kinetics associated with the reactions leading tometal deposition. It is conceivable that the length of thepulses is comparable in time with the kinetics of thedeposition reaction — this is especially true for cobaltwith acetate as the complexing agent. Although acetateis a relatively weak complexing agent for both nickeland cobalt (log K1 is 1.12 for nickel and 1.5 for cobalt),and the huge amount of chloride also acts as complex-ing agent, the addition of acetate does change thedeposition potential thus influencing the electrochemi-cal reactions.

The amount of cobalt in the deposit depends pre-dominately on the amount of cobalt in the electrolyte(see Fig. 5). A bath concentration of 2.0 g/l cobalt givesfrom 12 to 15 at.% Co in the deposit, and a concentra-tion of 4.0 g/l of cobalt leads to a deposit content from22 to 25% Co. The average current density and thepresence of complexing agents (acetate) seems less im-portant with respect to alloy composition, but DCplating increases the cobalt contents compared to PRfor the same electrolyte. This observation could also beexplained by the fact that deposition of cobalt is re-garded as a relatively slow process compared to nickel,and as such the deposition of cobalt will be moreinfluenced by the reversing current than nickel.

The NiCo electrolytes were very similar to the purenickel bath, and with a current efficiency around 94%.The as-plated residual stress was 40 MPa.

The hardness of the deposited NiCo alloy (with 15at.% Co) was measured to 380 HV100 g, which is muchhigher than for pure nickel. Deposits with approxi-mately 30 at.% of cobalt are expected to be the hardest,and future work will continue in this direction.

3. Results and discussion

Some properties of PR plated pure nickel, such ascorrosion resistance and texture [4], and residual stressas a function of current density [5] have previously beendescribed. The electrolyte, as described in Table 1, ischemically very stable with a current efficiency ofnearly 100%. The hardness of PR plated deposits isaround 220 HV100 g and the as-plated residual stress 80MPa (by cooling the plated film on the silicon substratedown from 40 °C to room temperature, additionalthermal stress should be added to the as-plated stress)(Fig. 2).

Fig. 7. Co (�), Fe (�) and Ni (�) in the deposit as a function ofaverage current density for PR plated CoNiFe alloys. The smallsymbols indicate experiments with copper foils as substrate (series I inFig. 6), the large symbols are plated from the electrolyte listed inTable 1 with silicon wafers as substrate (series IV in Fig. 6).

P.T. Tang / Electrochimica Acta 47 (2001) 61–66 65

Fig. 8. Micromechanical mirror fabricated in NiCo (8 �m thick with 15 at.% Co). The mirror is suspended by two beams, and can be moved byelectrostatic forces using electrodes under the mirror (contacted by the rectangular contact pads). The holes in the mirror surface enable thesolution for selective etching of the sacrificial layer to work faster. In the SEM micrograph, the sacrificial layer is only partly dissolved revealingthe plating base. The fabrication sequence is explained in detail for pure nickel structures by Johansen [8] and others.

3.2. CoNiFe

In this investigation CoNiFe deposits on siliconwafers (series IV in Fig. 6) was measured by a vibratingsample magnetometer. A deposit with 58.5 at.% Co,18.1 at.% Ni and 23.4 at.% Fe (and no sulphur) exhib-ited a Bs of 1.7 T and a coercivity of 510 A/m (a littlemore than 6 Oe) [7]. Since the naphthalene 1,3,6-trisul-phonic acid used in this investigation does not lead toco-deposition of sulphur (at least not in the concentra-tions used), it is likely to assume that the alloy compo-sitions of series IV are to far away from the phaseboundary to have the optimal microstructure for themagnetic properties (Fig. 7 Fig. 8)).

The CoNiFe electrolytes contain 5-sulfosalicylic acid,which is a very strong complexing agent for Fe3+

(log K1 is 14.64), and the bath seems very stable withoutany signs of Fe3+-precipitates. The current efficiency is96% and the residual stresses only 25 MPa. Thesevalues are much better than the electroplated PermAl-loy typically operating with current efficiencies of 70%or lower [7] and residual stresses of more than 100 MPa[7] — even with saccharin as the stress reducer.

Future experiments with the CoNiFe alloys will focuson increasing the saturation flux density and decreasingthe coercivity — primarily by microstructure analysisof various alloy compositions. The obtained magneticproperties are promising, since Bs is much higher thanfor PermAlloy, and the alloy could be used in applica-tion where a very low coercivity is not essential.

4. Conclusions

The advantages of using PR plating for microgal-vanics are:• reduced tendency to build up thick deposits in tradi-

tional high current density areas and improved stepcoverage without pores reaching down to thesubstrate;

• excellent bath stability with almost 100% currentefficiency and no additive consumption;

• good control of residual stress without the use ofadditives leading to co-deposition of sulphur;

• very low tendency to form pitting even withoutwetting agents.The disadvantages are:

• more expensive power supplies;• deposits without residual stress have not been

obtained;• very corrosive electrolytes because of the high chlo-

ride contents;• deposition rates higher than 50 �m/h are difficult to

obtain without risking passivation.

Acknowledgements

The author wishes to thank Frank Rasmussen andJan Tue Ravnkilde of Mikroelektronik Centret (MIC)at the Technical University of Denmark, for extendedassistance with deposition and characterisation of the

P.T. Tang / Electrochimica Acta 47 (2001) 61–6666

CoNiFe alloys as well as lithography work on thepresented MEMS structure.

References

[1] O. Geschke, W. Rong, P.T. Tang, J.P. Kutter, P. Telleman,Polymer Structures for �Tas, EuroSensors, Copenhagen, 27–30August 2000.

[2] W. Paatsch, Galvanotechnik mit Strompulsen—Teil 1: Nickelab-scheidung, Metalloberflache 40 (9) (1986) 387.

[3] G.W. Jernstedt, Better Deposits at Greater Speeds by PR Plating,Plating (reprint 4404) July, 1948.

[4] P.T. Tang, T. Watanabe, J.E.T. Andersen, G. Bech-Nielsen,Improved corrosion resistance of pulse plated nickel through

crystallisation control, Journal of Applied Electrochemistry 25(1995) 347.

[5] P.T. Tang, H. Dylmer, P. Møller, An Electroplating Method ofForming Platings of Nickel, Cobalt, Nickel Alloys and CobaltAlloys with Reduced Stress, EP 0835335 and US 6,036,833.

[6] T. Osaka, M. Takai, K. Hayashi, K. Ohashi, M. Saito, K.Yamada, A soft magnetic CoNiFe film with high saturationmagnetic flux density and low coercivity, Nature 392 (1998) 796.

[7] F.E. Rasmussen, J.T. Ravnkilde, P.T. Tang, O. Hansen, S. Bouw-stra, Electroplating and Characterisation of Cobalt–Nickel–Ironand Nickel–Iron for Magnetic Microsystem Applications, Eu-roSensors, Copenhagen, 27–30 August 2000.

[8] L.S. Johansen, M. Ginnerup, P.T. Tang, J.T. Ravnkilde, B.Lochel, Fabrication of Electroplated 3D MicrostructuresCombining KOH Etching, Electrodeposition of Photoresist andSelective Etching, Transducers, Sendai, 7–10 June 1999 .

.