`
`Journals About Contact us My IOPscience
`
`Development of new electrolytic and electroless gold plating processes for electronics
`
`applications
`
`This content has been downloaded from IOPscience. Please scroll down to see the full text.
`
`2006 Sci. Technol. Adv. Mater. 7 425
`
`(http://iopscience.iop.org/1468-6996/7/5/A04)
`
`View the table of contents for this issue, or go to the journal homepage for more
`
`Download details:
`
`IP Address: 108.6.17.84
`This content was downloaded on 26/04/2015 at 10:21
`
`Please note that terms and conditions apply.
`
`IPR2015-01087 - Ex. 1037
`Micron Technology, Inc., et al., Petitioners
`1
`
`
`
`ARTICLE IN PRESS
`
`Science and Technology of Advanced Materials 7 (2006) 425–437
`
`www.elsevier.com/locate/stam
`
`Review
`
`Development of new electrolytic and electroless gold plating processes
`for electronics applications
`Tetsuya Osakaa,b,c, , Yutaka Okinakab, Junji Sasanoc, Masaru Katod
`
`aDepartment of Applied Chemistry, School of Science and Engineering, Waseda University, Tokyo 169-8555, Japan
`bAdvanced Research Institute for Science and Engineering, Waseda University, Tokyo 169-8555, Japan
`cKagami Memorial Laboratory for Materials Science and Technology, Waseda University, Tokyo 169-0051, Japan
`dCentral Research Laboratory, Kanto Chemical Co. Inc., Saitama 340-0003, Japan
`
`Received 27 December 2005; received in revised form 28 April 2006; accepted 8 May 2006
`Available online 18 July 2006
`
`Abstract
`
`This article reviews results of our investigations, performed over the period of a decade, on gold plating for electronics applications.
`Three different topics are covered: (1) development of a new, non-cyanide, soft-gold electroplating bath containing both thiosulfate and
`sulfite as ligands; (2) evaluation of a known cyanide-based, substrate-catalyzed electroless bath for depositing pure soft gold, and
`subsequent development of an alternative, non-cyanide, substrate-catalyzed bath; and (3) development of a new process to electroplate
`amorphous hard-gold alloys for probable future applications as a contact material on nano-scale electronic devices.
`r 2006 NIMS and Elsevier Ltd. All rights reserved.
`
`Keywords: Gold; Electroplating; Electroless plating; Substrate-catalyzed plating; Amorphous gold alloys
`
`Contents
`
`Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
`1.
`2. Non-cyanide thiosulfate–sulfite bath for electroplating soft gold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
`3.
`Substrate-catalyzed electroless (SCEL) plating of soft gold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
`3.1. Cyanide bath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
`3.1.1. Ni–B vs. Ni–P as substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
`3.1.2.
`Improving the uniformity and adherence of gold deposit on Ni–P. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
`3.2. Non-cyanide bath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
`3.2.1.
`Identification of the mechanism of gold deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
`3.2.2. Gold deposits formed on Ni–B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
`3.2.3. Gold deposits formed on Ni–P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
`4. Electroplating of amorphous gold alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
`4.1. Effect of the addition of KAu(CN)2 in the Ni–W bath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
`4.2. Physical properties of amorphous gold–nickel alloy film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
`4.3. Amorphous Au–Co alloy film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
`Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
`Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
`References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
`
`5.
`
` Corresponding author. Department of Applied Chemistry, School of Science and Engineering, Waseda University, Tokyo 169-8555, Japan.
`Tel.:+81 3 5286 3202; fax:+81 3 3205 2074.
`E-mail address: osakatet@waseda.jp (T. Osaka).
`
`1468-6996/$ - see front matter r 2006 NIMS and Elsevier Ltd. All rights reserved.
`doi:10.1016/j.stam.2006.05.003
`
`The STAM archive is now available from the IOP Publishing website
`
`http://www.iop.org/journals/STAM
`
`2
`
`
`
`426
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`ARTICLE IN PRESS
`
`1. Introduction
`
`Both electrolytic and electroless gold plating processes
`are indispensable for the fabrication of electrical contacts
`in the electronics industry. The materials of plated gold in
`use for this purpose can be classified into two categories:
`soft gold and hard gold. Soft gold is used for metallizing
`bonding pads and fabricating microbumps on silicon IC
`chips and ceramic packaging boards, while hard gold is
`used as a contact material on electrical connectors, printed
`circuit boards, and mechanical relays. Electrolytic methods
`are available for plating both soft gold and hard gold,
`whereas electroless methods can produce only soft gold at
`present. In this article, our past contributions to the
`development of processes for electro- and electroless-
`plating of soft gold and electroplating of amorphous
`hard-gold alloys are reviewed in three sections (Sections 2,
`3, and 4).
`Section 2 summarizes results of our investigation carried
`out to formulate a new non-cyanide, soft-gold electroplat-
`ing bath containing both thiosulfate and sulfite as ligands
`for Au(I) [1,2]. This bath is highly stable, and unlike the
`conventional sulfite bath, it does not require the addition of
`a stabilizing agent.
`Section 3, which consists of Sections 3.1 and 3.2,
`describes our investigation of substrate-catalyzed electro-
`less (hereafter abbreviated as SCEL) processes for deposit-
`ing pure, soft gold, as distinguished from the conventional
`galvanic displacement and autocatalytic processes. The
`SCEL process is of interest because of its two distinct
`advantages over the other conventional electroless pro-
`cesses: (1) the gold film obtained is much less porous than
`that deposited in the galvanic displacement bath, and (2)
`the SCEL bath is much more stable than the autocatalytic
`bath. The original SCEL bath was developed by Iacov-
`angelo and Zarnoch [3] in 1991. We evaluated this bath in
`great detail to understand its general characteristics and
`properties of the gold deposit obtained [4], and the results
`are summarized in Section 3.1. Subsequently, we extended
`this work to develop a non-cyanide SCEL bath containing
`thiosulfate and sulfite as ligands [5,6]. The work on the
`non-cyanide SCEL bath is reviewed in Section 3.2. Non-
`cyanide baths are always preferred because they are non-
`toxic and more likely to be compatible with conventional
`positive photoresists employed to delineate electronic
`circuit patterns.
`Section 4 of this article reviews results of our more recent
`attempt to develop a process for electroplating amorphous
`gold alloys. This work was initiated with the aim of
`creating an electroplated hard-gold film suitable as an
`electrical contact material on submicro- and nano-scale
`electronic devices. Critical mechanical properties, such as
`hardness and wear resistance, of the conventional crystal-
`line hard-gold films result from their small grain size, of the
`order of 20–30 nm. Therefore, when the size or the physical
`dimensions of the contact surface becomes comparable to
`or even smaller than the grain size in nano-devices of the
`
`next generation, the properties of hard-gold films of such
`small dimensions are expected to deviate significantly from
`those of bulk hard gold. On the other hand, mechanical
`properties of amorphous metals and alloys in general are
`known to be independent of their size because of the
`absence of crystal grains in such materials. Thus, to be
`prepared for the future need of an amorphous electrical
`contact material, we made a preliminary investigation on
`the possibility of electroplating amorphous Au–Ni alloy
`from a bath prepared by adding a gold salt into a bath that
`is already known to deposit an amorphous alloy such as
`Ni–W [7]. This approach was successful in developing a
`process to electroplate amorphous Au–Ni alloy with a
`hardness value greater than twice that of the conventional
`hard gold without adversely affecting the electrical contact
`resistance [8,9]. Using a similar approach, we also
`developed a process for electroplating amorphous Au–Co
`alloy films [9].
`Our contributions to the development of these new
`plating processes are reviewed below.
`
`2. Non-cyanide thiosulfate–sulfite bath for electroplating
`soft gold
`
`Electroplating of soft gold is generally carried out with a
`bath containing KAu(CN)2 in a phosphate buffer of pH 7
`at a mildly elevated temperature. It does not contain free
`cyanide ions initially, but they are generated at the cathode
`surface as a result of the gold deposition reaction. The
`cyanide ions are partly converted into HCN, which escape
`into the atmosphere, and partly remain in the bath. The
`presence of free cyanide is undesirable not only for its
`toxicity but also for its incompatibility with photoresists
`used to delineate circuit patterns through which the gold is
`to be plated. Cyanide attacks the interface between the
`substrate and the photoresist,
`lifting the latter and
`depositing extraneous gold under the photoresist. To avoid
`these problems, an Au(I) sulfite bath is used instead.
`However, the sulfite bath easily undergoes a disproportio-
`nation reaction to form Au(III) and metallic Au because of
`the relatively low stability of the Au(I) sulfite complex,
`causing the bath to decompose spontaneously on standing,
`unless a suitable stabilizer is added to the bath. All
`commercially available Au(I) sulfite baths contain proprie-
`tary stabilizing additives.
`In contrast, the Au(I) thiosulfate–sulfite mixed ligand
`bath we developed is highly stable and requires no
`stabilizing additive [1,2]. We selected this system in view
`of the fact that it is used successfully for formulating non-
`cyanide, autocatalytic electroless gold plating baths yield-
`ing good bath stability and deposit properties [10–12].
`The composition and operating conditions of the bath
`optimized for obtaining gold deposits with the lowest
`possible hardness are shown in Table 1. Because the aim of
`developing this bath was to apply it to the fabrication of
`gold microbumps to be used for attaching IC chips to
`circuit packages, the hardness of the gold was desired to be
`
`3
`
`
`
`ARTICLE IN PRESS
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`427
`
`Sulfur content in deposit [ppm]
`
`200
`
`100
`
`2
`
`1
`
`10
`20
`Tl+ concentration [ppm]
`
`0
`
`30
`
`120
`
`100
`
`80
`
`Vickers hardness [kg mm-2]
`
`Table 1
`Compositions and operating conditions of thiosulfate–sulfite electroplat-
`ing baths and Vickers hardness values of gold deposits obtained [1]
`
`Reagent
`
`Without Tl+
`
`With Tl+
`
`Bath A
`
`Bath B
`
`Bath C
`
`NaAuCl4 2H2O
`Na2SO3
`Na2S2O3
`Na2HPO4
`Tl+ (added as Tl2SO4)
`pH
`Temperature
`Current density
`Rotation speed (Disk electrode)
` 2)
`Hardness (kg mm
`112
`As deposited
`After annealing at 350 1C for 30 min 100
`
`0.06 M
`0.06 M
`0.06 M
`1.1–1.4 M 0.42 M
`0.42 M
`1.1–1.4 M 0.42 M
`0.42 M
`0.30 M
`0.30 M
`0.30 M
`0
`5 ppm
`0
`6.0
`6.0
`6.0
`60 1C
`60 1C
`60 1C
`
`
`5 mA cm
`500 rpm 500 rpm 500 rpm
`
` 2 2 5 mA cm 2 5 mA cm
`
`87–88
`45–60
`
`88
`52
`
`60
`
`0
`
`Fig. 1. Effects of thallium ion concentration on hardness (curve 1) and
`sulfur content (curve 2) of gold deposits obtained in thiosulfate–sulfite
`electroplating bath [1]. (For concentrations of other bath constituents and
`operating conditions, see Table 1.)
`
`Sulfur content of gold deposit [ppm]
`
`160
`
`120
`
`80
`
`40
`
`1
`
`2
`
`120
`
`100
`
`Vickers hardness [kg mm-2]
`
`80
`
`0.8
`
`1.4
`2.0
`2.6
`Total concentration of Na2S2O3 and NaSO3 [M]
`
`Fig. 2. Effect of total ligand concentration on hardness (curve 1) and
`sulfur content (curve 2) of gold deposits obtained in thiosulfate–sulfite
`electroplating bath [1]. ([Na2S2O3]:[Na2SO3] ¼ 1:1). (For concentrations
`of other bath constituents and operating conditions, see Table 1.)
`
`metallized with evaporated thin layers of Al/Ti/W/Au.
`The bump patterning was carried out using a conventional
`photoresist. A scanning electron microscopy (SEM)
`examination of the bumps after removal of the photoresist
`proved that there was no extraneous gold deposition and
`that sidewalls of the bumps were straight, indicating that
`the photoresist withstood without degradation during the
`gold deposition process [2].
`
`(Bath A, low ligand concentration; Bath B, high ligand concentration;
`Bath C, low ligand concentration with Tl+ addition).
`
`as low as possible. Our investigation revealed that two
`approaches, besides annealing, are effective for achieving
`this purpose. The first approach is to add Tl+ ions at a
`concentration as small as 5 ppm to the bath. This method
`was found to decrease the deposit hardness from 112 to
` 2 in Vickers hardness (compare Bath A and Bath
`88 kg mm
`C in Table 1). For the conventional cyanide bath, Tl+ is
`known as a grain refiner. Fig. 1 demonstrates the effect of
`thallium concentration on the hardness of the gold deposit
`formed in the thiosulfate–sulfite bath [1]. The second
`approach we found effective for decreasing the hardness is
`to increase the thiosulfate concentration. The result shown
`in Fig. 2 was obtained by varying the total concentration of
`thiosulfate and sulfite, while the concentration ratio of the
`two ions was kept equal to unity. Hence, the thiosulfate
`concentrations were equal to one half of the values of total
`ligand concentration shown on the horizontal axis. In a
`separate experiment in which the thiosulfate concentration
`was varied independently from the sulfite concentration,
`the observed decrease in hardness was shown to be due to
`the increase in thiosulfate concentration [2].
`We carried out a detailed study for understanding the
`reason why the deposit hardness is affected by the two
`variables of bath composition described above. It was
`found that
`the observed decrease in hardness closely
`parallels the decrease in sulfur content of the deposit.
`The sulfur content is also plotted in Figs. 1 and 2. Our
`study also showed that the origin of the included sulfur is
`primarily the adsorbed Au(I) thiosulfate species in the form
`of (AuS2O3)ads formed as an intermediate in the gold
`deposition reaction from [Au(S2O3)2]3
`present
`in the
`bath [2].
`To demonstrate the practical usefulness of the thiosul-
`fate-sulfite bath, approximately 30-mm-thick bumps mea-
`suring 50 50 mm were
`formed on a silicon wafer
`
`4
`
`
`
`428
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`ARTICLE IN PRESS
`
`3. Substrate-catalyzed electroless (SCEL) plating of soft
`gold
`
`Three fundamentally different electroless gold plating
`processes are known: (1) galvanic displacement process, (2)
`autocatalytic process, and (3) substrate-catalyzed (SCEL)
`process. We found that the SCEL process is advantageous
`over the galvanic displacement process in that the deposit
`obtained by the former method is significantly less porous
`than that produced by the latter. [3,4] Compared to the
`autocatalytic bath, the SCEL bath is more stable and less
`susceptible to spontaneous decomposition, although the
`maximum gold thickness obtainable with the latter bath is
`limited. In spite of these advantages, the SCEL process had
`been studied to a much lesser extent than the other two
`types of electroless methods. Our investigation of the
`SCEL processes described below was initiated under these
`circumstances.
`
`3.1. Cyanide bath
`
`The original SCEL bath described by Iacovangelo and
`Zarnoch [3] contained KAu(CN)2, KCN, KOH, and
`K2CO3 with hydrazine as the reducing agent (Table 2).
`Using this original bath, we investigated effects of the
`composition and the pretreatment procedure of various
`electroless nickel substrates on the uniformity and adher-
`ence of the gold deposits produced. The reason why
`electroless nickel substrates were chosen for this study was
`that they are most commonly used as the barrier material
`to prevent diffusion of copper atoms from the underlying
`copper substrate through the gold film.
`
`3.1.1. Ni–B vs. Ni–P as substrate
`The SCEL gold deposited on electroless Ni–B (5.4 wt%
`B) substrate was found to be invariably adherent and
`uniform in appearance, whereas the same deposit formed
`on electroless Ni–P (various P contents) was non-adherent
`and non-uniform regardless of the P content. The only
`exception was the low P-content substrate pretreated with a
`specific solution described in the subsequent section. Fig. 3
`compares SEM photographs of SCEL gold plated for
`various lengths of time on electroless Ni–B (5.4 wt% B) and
`Ni–P (15.4 wt% P)
`substrates, both pretreated with
`
`Table 2
`Composition and operating conditions of cyanide-based substrate-
`catalyzed electroless gold plating bath [3]
`
`Reagent
`
`K2CO3
`KOH
`KCN
`Kau(CN)2
`N2H4 H2O
`Agitation
`Temperature
`
` 3)
`Concentration (mol dm
`
`0.75
`0.87
`0.01
`0.017
`0.50
`Magnetic stirrer
`80 1C
`
`10 vol% HCl. On the Ni–B substrate, numerous crystals
`of uniform sizes were observed after the initial plating
`period of only 5 s (Fig. 3–a-1), and after 30 s the substrate
`was covered uniformly and completely with fine gold
`crystals (Fig. 3-a-2 and a-3). On the other hand, only a
`small number of gold crystals were observed on the Ni–P
`substrate even after 30 s (Fig. 3-b-1). The photograph of
`Fig. 3-b-2 was taken after 3 min of deposition time, which
`shows that the crystals observed at 30 s grew in size, while
`the number density of crystals remained essentially un-
`changed between 30 s and 3 min, indicating that no further
`nucleation took place during that period. Fig. 3-b-3, taken
`after 5 min, shows that the crystals grew further in size and
`coalesced to form agglomerates. These results indicate that
`the uniformity and adherence of the gold deposits are
`related to the density of gold nuclei produced at the initial
`stages of gold deposition.
`
`3.1.2. Improving the uniformity and adherence of gold
`deposit on Ni–P
`the significantly better uniformity and
`Because of
`adherence of the gold deposited on Ni–B, this material
`would be preferred to Ni–P as the substrate for gold
`deposition from the SCEL bath. However, electroless Ni–P
`is considered more desirable than Ni–B for practical
`reasons such as the better bath stability, the greater ease
`of process control, and the lower cost. Therefore, a series of
`investigations were carried out to find out whether the
`uniformity and adherence of gold deposits formed on
`electroless Ni–P can be improved.
`To investigate the effect of phosphorus content, Ni–P
`deposits with high, medium, and low P contents were
`prepared by varying bath pH [13]. These substrates were
`subjected to pretreatment with either one of the two
`solutions: (a) 10 vol% HCl or (b) a mixture of 0.1 M
`NH4F and 0.1 M sodium sulfamate (designated as FS
`mixture) [4]. The FS treatment was performed by immer-
`sion in the above mixture for 15 s at 70 1C. Non-adherent,
`non-uniform deposits were obtained on both high
`(15.4 wt%) and medium (10.5 wt%) phosphorus content
`substrates regardless of which solution was used for
`pretreatment, whereas an adherent and uniform gold
`deposit was obtained successfully on the low phosphorus
`content Ni–P (4.7 wt% P) provided that it was pretreated by
`the FS mixture. Fig. 4 illustrates the effect of P content of
`electroless Ni–P substrates pretreated with the FS mixture
`on the morphology of gold deposits produced after various
`deposition times. It is seen that the nucleation density is
`much greater on the low P (4.7 wt% P) substrate (Fig. 4-a-1
`to a-3) than on the high P (15.4 wt% P) substrate (Fig. 4-b-1
`to b-3). On the latter substrate, the nuclei scattered on the
`surface grew in size with time, but even after 5 min, the
`coverage was incomplete.
`Both HCl and FS pretreatment methods are considered
`to remove passive oxide films on the surface of Ni–P. It was
`surprising that the substrate-catalyzed method did not
`yield acceptable gold deposits on the Ni–P substrates with
`
`5
`
`
`
`ARTICLE IN PRESS
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`429
`
`Fig. 3. SEM images of gold deposits formed in cyanide-based substrate-catalyzed bath on two different nickel substrates: (a) Ni–B (5.4 wt% B) and (b)
`Ni–P (15.4 wt% P). Plating time: (a-1) 5 s, (a-2) 30 s, and (a-3) 3 min; (b-1) 30 s, (b-2) 3 min, and (b-3) 5 min [4].
`
`high and medium P contents even after these pretreatment
`processes. Apparently, the passive oxide film naturally
`formed on the high and medium P-content Ni–P substrates
`was so highly inert
`that even the FS treatment was
`ineffective for its removal, whereas the pretreatment easily
`attacked the passive film on the low-P-content substrate. It
`was thus concluded that if electroless Ni–P is to be used
`instead of Ni–B as the substrate to plate gold using the
`cyanide-based SCEL bath, a low-P-content substrate must
`be chosen and its surface treated with a suitable activator
`such as the FS mixture.
`
`3.2. Non-cyanide bath
`
`In spite of the advantages described above, the presence
`of a high concentration of free cyanide and the high
`alkalinity of the cyanide SCEL bath are undesirable for the
`reasons mentioned already. Thus, we attempted to
`formulate a non-cyanide [5,6] SCEL bath containing
`thiosulfate and sulfite as ligands for Au(I).
`The composition and operating conditions of the non-
`cyanide SCEL bath are listed in Table 3, which are similar
`to those of the autocatalytic bath containing the same
`ligands [10] except that the reducing agent (ascorbic acid)
`was excluded. For bath makeup, the commonly available
`
`trivalent gold salt, NaAuCl4, is used, but the Au(III) in this
`salt is reduced immediately to Au(I) upon addition of
`Na2SO3, forming Au(I) sulfite complex. This complex is
`converted to Au(I)
`thiosulfate complex and/or Au(I)
`([Au(S2O3)
`thiosulfate–sulfite mixed ligand complex
`(SO3)]3
`) upon addition of Na2S2O3 because of the greater
`stability of those complexes than the sulfite complex.[11,12]
`We have shown by electrochemical polarization measure-
`ments [6] that in this SCEL system, sulfite ions serve
`as the reducing agent. The composition of the galvanic
`displacement bath used for comparison is shown in
`Table 4.
`
`3.2.1. Identification of the mechanism of gold deposition
`Fig. 5 compares the two substrates, electroless Ni–B film
`and pure gold sheet, for the variation of gold deposit
`thickness with plating time in the non-cyanide SCEL bath.
`Gold deposition took place only on Ni–B and not on the
`gold sheet. This
`result clearly shows
`that
`the gold
`deposition reaction is not autocatalytic. To find out
`whether galvanic displacement is involved in the gold
`deposition on Ni–B, the concentration of dissolved nickel
`ions was determined after immersion of the substrate in
`100 mL of the bath for 60 min. It was found to be 0.99 ppm,
`which is equivalent to 0.04 mm in gold thickness, whereas
`
`6
`
`
`
`430
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`ARTICLE IN PRESS
`
`Fig. 4. SEM images of gold deposits formed in cyanide-based substrate-catalyzed bath on low- and high-P-content Ni–P substrates: (a) Ni–P (4.7 wt% P)
`and (b) Ni–P (15.4 wt% P) pretreated with FS mixture. Plating time: (a-1) 5 s, (a-2) 30 s, and (a-3) 3 min; (b-1) 30 s, (b-2) 3 min, and (b-3) 5 min [4].
`
`Table 3
`Composition and operating conditions of non-cyanide substrate-catalyzed
`gold plating bath [5]
`
`Reagent
`NaAuCl4 2H2O
`Na2SO3
`Na2S2O3 5H2O
`Na2HPO4
`pH (adjusted with NaOH)
`Agitation
`Temperature
`
` 3)
`Concentration (mol dm
`
`0.01
`0.32
`0.08
`0.32
`9.0
`Mechanical stirrer
`60 1C
`
`Table 4
`Composition and operating conditions of galvanic displacement gold
`plating bath [5]
`
`Reagent
`
`Na3Au(SO3)2
`Na2SO3
`Citric acid
`pH (adjusted with H2SO4)
`Temperature
`
` 3)
`Concentration (mol dm
`
`0.05
`0.54
`0.14
`7.0
`851 C
`
`the actual gold thickness measured was as large as 0.21 mm.
`Thus, the galvanic displacement reaction accounted for
`approximately 19% of the total deposit thickness, and
`
`there was another mechanism by which the remaining 81%
`of the gold was deposited. This mechanism is accounted for
`by the substrate-catalyzed reaction. Fig. 6 compares the
`calculated and actual gold thicknesses of
`the deposit
`formed in the bath of Table 3 and those of the deposit
`produced in the galvanic displacement bath of Table 4. It is
`seen that for the latter bath, the gold thickness calculated
`from the amount of dissolved nickel agrees well with the
`actually determined thickness as expected.
`For practical applications, it is important to ascertain
`that the process produces gold deposits with acceptable
`appearance, uniformity, adherence, and low or preferably
`no porosity. Also from the practical standpoint,
`it is
`desirable to understand effects of various variables such as
`bath composition and substrates on the above attributes.
`We conducted a systematic study of those effects with
`substrates of not only electroless Ni–B film but also
`electroless Ni–P films with high and low P contents.
`
`3.2.2. Gold deposits formed on Ni–B
`As mentioned in Section 3.1, the cyanide-based SCEL
`bath always yielded adherent and uniform gold deposits on
`the Ni–B substrate pretreated by simple immersion in
`10 vol% HCl. In contrast, the non-cyanide, thiosulfate–
`sulfite bath with the standard composition shown in
`
`7
`
`
`
`ARTICLE IN PRESS
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`431
`
`composition, effects of thiosulfate and sulfite concentra-
`tions were investigated. The results are illustrated in Fig. 7.
`In this series of experiments, the Ni–B substrates were
`pretreated only by room-temperature
`immersion in
`10 vol% HCl. The deposit uniformity clearly improved as
`the thiosulfate concentration was decreased from 0.08 to
`0.01 M at all three sulfite concentrations of 0.08, 0.16, and
`0.32 M. This observation points to the speculation that
`thiosulfate is responsible for the deposit non-uniformity
`observed more prominently at
`the higher thiosulfate
`concentrations. In an attempt to clarify the cause of the
`observed non-uniformity, Auger electron spectroscopy
`(AES) was performed on a specimen with a non-uniform
`gold film deposited on Ni–B [4]. The AES analysis of the
`Ni–B surface exposed after peeling off the gold film from a
`blistered area showed the presence of S in addition to O
`and C, whereas the AES depth profile of an adherent
`portion of the gold film did not reveal the presence of S.
`These AES results appear to indicate that the localized
`presence of an S-containing species is responsible for the
`loss of adherence of the gold film to the Ni–B substrate.
`
`3.2.3. Gold deposits formed on Ni–P
`As was the case with the cyanide-based system [4], it was
`not possible to produce an adherent and uniform gold
`deposit on the high P content (15.4 wt% P) Ni–P substrate.
`The introduction of a pretreatment procedure using either
`10 vol% HCl, the FS mixture, or 0.1 M Na2SO3, did not
`help improve the deposit characteristics. Modification of
`the bath composition did not result in any improvement.
`The reason for this behavior of the high P content Ni–P is
`believed to be the presence of a highly stable, tenacious
`passive oxide film and the inability of the simple pretreat-
`ment procedures to remove such a film.
`In contrast to the high P content Ni–P, the low P content
`(4.7 wt% P) Ni–P substrate pretreated with 0.1 M Na2SO3
`yielded a very uniform, adherent gold film at the thiosulfate
`concentration of 0.01 M. Even when the thiosulfate
`concentration was increased to the level of the basic bath
`(0.08 M), the deposit uniformity was good. Fig. 8 shows the
`variation of gold thickness with deposition time at three
`different
`thiosulfate concentrations. Concentrations of
`other bath constituents and bath operating conditions
`were the same as those for the basic bath (Table 3). It is
`seen that the gold thickness reaches a maximum constant
`value after approximately 60 min of plating at all thiosul-
`fate concentrations used, and that the maximum thickness
`value increases with increasing thiosulfate concentration. A
`similar dependence of the maximum gold thickness on
`ligand concentration is known for the cyanide-based
`system [3,4]. The photographs pasted in the large squares
`on both sides of the graph in Fig. 8 show the appearance of
`the specimens tested for porosity. The dark spots and areas
`in each specimen show the presence of colored Ni-
`dimethylglyoxime complex formed in the test from exposed
`Ni metal through the pores in the gold film. The actual
`color of
`these spots and areas was pink. The three
`
`On Ni-B
`
`On Au
`
`20
`
`40
`
`60
`
`Time [min.]
`
`0.4
`
`0.3
`
`0.2
`
`0.1
`
`0
`
`Gold thickness [µm]
`
`Fig. 5. Comparison of variations of deposit thickness with time on Ni–B
`and Au substrates for non-cyanide substrate-catalyzed bath [5].
`
`Substrate-catalyzed
`deposition
`
`0.25
`
`0.20
`
`0.15
`
`0.10
`
`0.05
`
`Gold thickness [µm]
`
`(a)
`
`0
`
`Calculated Measured
`
`Calculated
`
`Measured
`
`(b)
`
`Fig. 6. Comparison between calculated and measured thicknesses of gold
`deposits produced in (a) non-cyanide substrate-catalyzed bath and (b)
`galvanic displacement bath. (Calculated thicknesses were estimated from
`the amount of dissolved Ni. Plating time was 60 min.) [5].
`
`Table 3 yielded gold deposits with non-uniform appearance
`on Ni–B. For the cyanide-based system, the substrate
`pretreatment in the FS mixture at 70 1C was highly effective
`for producing gold deposits with uniform and bright
`appearance, as described in the preceding section. For the
`non-cyanide SCEL system, this pretreatment procedure
`was found to be only slightly more effective than the HCl
`treatment for improving the deposit uniformity. Unlike the
`case of the low P content Ni–P substrate described in the
`subsequent section, the pretreatment with 0.1 M Na2SO3
`did not improve the uniformity of the deposit produced on
`the Ni–B substrate. To learn whether the deposit uni-
`formity can be improved by manipulating the bath
`
`8
`
`
`
`432
`
`T. Osaka et al. / Science and Technology of Advanced Materials 7 (2006) 425–437
`
`ARTICLE IN PRESS
`
`Na2SO3
`
`Na2S2O3
`
`0.08 M
`
`0.16 M
`
`0.32 M
`
`0.01 M
`
`0.02 M
`
`0.04 M
`
`0.08 M
`
`10 mm
`
`Fig. 7. Effect of sulfite and thiosulfate concentrations on the appearance of gold deposits plated non-cyanide substrate-catalyzed bath on electroless Ni–B
`substrate. (Each substrate was pretreated in 10 vol% HCl.) [6].
`
`photographs on the right side of the figure represent
`specimens obtained after 60 min of deposition time at the
`three different thiosulfate concentrations. Remarkably, an
`essentially pore-free gold film was obtained at the thickness
`measuring only 0.05 mm when the thiosulfate concentration
`was 0.01 M. As the thiosulfate concentration was in-
`creased,
`the maximum gold thickness
`increased to
`0.28 mm at 0.24 M; nevertheless, the porosity also increased
`dramatically. The three specimens shown on the left side of
`Fig. 8 were obtained at the deposition time of 30 min. At
`0.24 M thiosulfate, the specimen obtained at 30 min was
`more porous than that obtained after 60 min. Comparison
`of the 30 min specimen prepared at 0.24 M thiosulfate with
`the ones made after 30 or 60 min at 0.08 M thiosulfate
`(specimens in the middle of both the right and the left rows)
`clearly shows that, despite the identical gold thickness of all
`three specimens, the film produced at the higher thiosulfate
`concentration was significantly more porous.
`To understand the mechanism determining the depen-
`dence of the maximum gold thickness and porosity on
`thiosulfate concentration, surface morphology was exam-
`
`ined using SEM for gold deposits produced at different
`deposition times in baths containing various