Electrochemical Copper Metallization of Glass Substrates Mediated by Solution-Phase Deposition of Adhesion-Promoting Layers

Metal-to-glass interfaces commonly encountered in electronics and surface ﬁnishing applications are prone to failure due to in- trinsically weak interfacial adhesion. In the present work, an ‘all-wet’ process (utilizing solution-phase process steps) is devel-oped for depositing nucleation- and adhesion-promoting layers that enhance the interfacial adhesion between glass substrates and electrochemically-deposited copper (Cu) ﬁlms. Adhesion between thick ( > 10 μ m) Cu ﬁlms and the underlying glass substrates is facilitated by an interfacial Pd-TiO 2 layer deposited using solution-phase processes. Additionally, the proposed interfacial engineer- ing utilizes self-assembled monolayers to functionalize the glass substrate, thereby improving surface wettability during Pd-TiO 2 deposition. Resulting Pd-TiO 2 deposits catalyze direct electroless plating of thin Cu seed layers, which enable subsequent electrode- position of thick ( > 10 μ m) Cu coatings. The present work provides a viable route for high-throughput, cost-effective metallization of glass and ceramic surfaces for electronics and surface ﬁnishing applications. the terms of unrestricted

The interfacial adhesion between metallic films and insulating substrates, e.g., glass, is intrinsically poor. This is a major roadblock in numerous electronics applications, particularly the manufacturing of printed circuit boards (PCBs) 1 and integrated circuits (ICs). 2,3 Interfacial adhesion between metallic thin films deposited on glass substrates can be improved using functionalized polymers 4 or selfassembled monolayers (SAMs). 2,5 SAMs provide interfacial 'anchoring' by chemically bonding to the substrate (SiO 2 ) as well as the deposited metal film. While SAMs have been shown to enable deposition of sub-micron scale, adherent Cu films on SiO 2 , 5 they do not provide adequate interfacial strength to enable thick (>10 μm) Cu coatings on glass. The Cu-to-SiO 2 interfacial adhesion may also be improved by utilizing metallic adhesion layers 6 or mixed-metal oxides. 7 Metallic adhesion promoters, e.g., titanium or titanium-nitride, are deposited using 'dry' methods such as physical vapor deposition (PVD). 'Dry' techniques are not desirable for high-volume manufacturing given their low throughput and high cost of ownership. On the other hand, mixed-metal oxides maybe deposited using sol-gel methods in which a metal alkoxide is co-deposited with a catalytic metal salt in a polarorganic solvent, resulting in a mixed-metal oxide adhesion layer with catalytic activity for electroless plating. 7 The methods by which the sol-gel catalyst may be deposited include dip-coating, printing, spincoating, or brushing.
In the present work, we demonstrate an 'all-wet' process for depositing adhesion-promoting layers that enhance the interfacial adhesion between glass substrates and electroless-deposited copper (Cu) films. Adhesion between electrochemically-deposited thick (>10 μm) Cu coatings and the underlying glass substrates is facilitated by an interfacial Pd-TiO 2 layer fabricated using solution-phase processing. Our interfacial engineering approach utilizes self-assembled monolayers (SAMs) to functionalize the glass substrates before applying the Pd-TiO 2 coating. The SAM layer improves surface wettability during Pd-TiO 2 deposition as shown below. Pd-TiO 2 deposits effectively catalyze electroless plating of thin Cu seed layers, which enable subsequent electrodeposition of thick (>10 μm) Cu coatings.

Experimental
This section describes the materials and methods used for cleaning the glass slides, functionalizing them with SAMs, depositing Pd-TiO 2 * Electrochemical Society Student Member.
* * Electrochemical Society Active Member. z E-mail: rna3@case.edu adhesion-promoting layers, depositing electroless and electroplated Cu films and characterization their adhesion.
Glass slide cleaning.-Eagle XG glass slides (alkaline earth boroaluminosilicate) manufactured by Corning were employed as substrates. Glass slides were cleaned using a protocol described by Cras. 8 First, glass slides were rinsed with deionized (DI) water and then dried using a stream of nitrogen. Next, the glass slides were submerged in a 1:1 volume ratio of concentrated hydrochloric acid and methanol for 30 min followed by a DI rinse. The glass slides were then submerged in concentrated sulfuric acid for 30 min followed by a thorough DI rinse and drying under nitrogen.
Silanization of cleaned glass surfaces.-After cleaning, glass slides were immersed in a 5 mM solution of APTES (3aminopropyltriethoxysilane, Acros Organics) in toluene solvent at 25 • C for 60 min. Slides were then rinsed with toluene, then with ethanol, and finally with DI water before drying under nitrogen.
Deposition of Pd-TiO 2 ink.-An ink solution containing 1.1 mM titanium (IV) butoxide (Acros) and 1.1 mM PdCl 2 in n-butanol solvent was prepared. The ink was dropped on the glass surface via a transfer pipette. Ink volume dropped was approximately 0.1 mL per 1 cm 2 of the glass surface area. The ink was then dried in air at 130 • C for 15 min in a Thermo Scientific Heratherm OGS 60 oven. Next, the dried ink was sintered in air at 450 • C in a Hoskins electric tube furnace for 30 min. After cool down, samples were immersed in a 2 M sulfuric acid solution for 2 min followed by a DI rinse. Finally, samples were immersed in a reductant solution of 0.5 M dimethylamine borane (DMAB) for 2 min and then rinsed with DI water.
We note that our aforementioned stepwise procedure of applying the Pd-TiO 2 adhesion-promoting layer differs from the sol-gel technique described elsewhere. 7 First, our procedure does not utilize chemical hydrolysis of the titanium butoxide in solution but rather relies on its reaction with moisture in air to induce oxidation (shown below). The DMAB reductant treatment then facilitates chemical reduction of the oxidized Pd to metallic Pd catalyst. Furthermore, we utilize dilute inks that enable thinner Pd-TiO 2 adhesion layers to be formed. Finally, our ink application procedure hinges on utilizing a SAM-terminated glass surface, which improves surface wettability and enables uniform Pd-TiO 2 deposition as also discussed below.
Electroless Cu deposition.-Proprietary electroless Cu plating chemistry from Atotech, USA was employed. This alkaline D631 electroless plating solution consisted of the following components: copper sulfate, tartrate-based complexing agent and formaldehydebased reducing agent. The electroless Cu deposition process was operated at 35 • C. After electroless plating, samples were rinsed with DI water and then annealed in a Hoskins electric tube furnace under flowing argon at 400 • C for 60 min.
Cu electroplating.-Following electroless seed layer deposition, samples were electroplated with Cu using a high-throw plating solution described by Dini and Snyder. 9 The composition of the bath was 100 g/L CuSO 4 .5H 2 O (Fisher Scientific), 270 g/L H 2 SO 4 (Fisher Scientific), and 0.1 g/L NaCl. A large area (70 cm 2 ) Cu foil counter electrode was used, which was placed at a distance of 10 cm from the working electrode. Samples were electroplated at a current density of 10 mA/cm 2 .
Materials characterization.-The scotch tape test was used to qualitatively assess the adhesion of electrochemically-deposited Cu on glass with or without the Pd-TiO 2 adhesion promoter layer. Scotch Matte Finish Magic tape from 3 M was used for all tape tests. During tape testing, the applied scotch tape was removed at roughly a 90 • angle relative to the sample. In some cases, prior to applying the tape, the Cu was scratched in perpendicular directions using a diamondtip scribe. For quantitative adhesion testing, a 90 • peel strength test was performed at a peel rate of 50 mm/min and the force required to peel unit width of the plated Cu was recorded. X-ray photoelectron spectroscopy (XPS) was performed using a PHI Versaprobe 5000 Scanning X-ray photoelectron spectrometer with a monochromatic AlKα anode X-ray source. Surface roughness of deposited Pd-TiO 2 layer was measured using a profilometer (KLA-Tencor P-6 Stylus) and surface morphology was imaged using a high-resolution optical microscope (Leica DM2500M). Figure 1 is a process flow diagram detailing the sequence of glass surface preparation, nucleation and adhesion promoter deposition, and Cu metallization used in the present work. Fig. 1a shows a cleaned, hydroxyl-terminated glass surface, which promotes the deposition of 3-aminopropytriethoxysilane (APTES) self-assembled monolayer (SAM) using procedure described above. APTES self-assembly is mediated by a condensation reaction 10 that provides amine-terminated glass as shown in Fig. 1b. APTES deposition plays an important role in improving the surface wettability, as discussed below.

Results and Discussion
Ink containing titanium butoxide and palladium chloride dissolved in n-butanol is then applied to the amine-terminated glass surface, following procedure described in the Experimental section above. Upon drying at 130 • C for 15 min, the ink leaves a composite mixture of PdCl 2 -TiO 2 on the glass surface (Fig. 1c). A prolonged (30 min), high temperature (450 • C) sintering step in air converts the PdCl 2 embedded in the TiO 2 matrix to PdO (Fig. 1d). This is supported by XPS data discussed below. Finally, the PdO is reduced to metallic Pd via reaction with 0.5 M dimethylamine borane (DMAB) for 2 min (Fig. 1e).
The Pd particles in the Pd-TiO 2 layer formed on the glass surface serve as catalytic sites for electroless Cu deposition. In our work, electroless deposition (using procedure outlined in Experimental section above) provided a ∼400 nm Cu seed layer on top of the Pd-TiO 2 adhesion promoting layer (Fig. 1f). Subsequently, Cu electrodeposition was performed on the electroless-seeded surface. Thick (>10 μm), adherent Cu electrodeposits were obtained (Fig. 1g). The glass/SAM/Pd-TiO 2 /Cu stack fabricated using the process flow shown in Fig. 1 was finally subjected to adhesion testing.
Wettability of APTES-modified glass surfaces. -Fig. 2 demonstrates the improved wettability of butanol ink on an APTES-modified glass slide. A 10 μL volume of ink (composition described in Experimental section above) was dropped onto the surface of APTESmodified glass slide. The droplet spread, which in a qualitative mea- sure of the surface wettability, was measured and compared to a glass slide without APTES termination. In the case of APTES-modified glass (Fig. 2a), the ink droplet spreads out nearly covering the entire surface of the glass indicating a low contact angle and good wettability. In absence of APTES (Fig. 2b), the ink droplet shows a higher contact angle and thus poor surface wetting. The improved wettability of the glass surface with APTES-termination is critical for the subsequent uniform deposition of the Pd-TiO 2 adhesion-promoting layer.
Pd-TiO 2 deposition mechanism.-To better understand the formation of the Pd-TiO 2 adhesion-promoting layer on the SAM-terminated glass surface, we performed XPS at various stages of ink application, drying, sintering and reduction. Fig. 3 shows the observed XPS spectra at each of these stages. Fig. 3a is XPS spectra of an APTES-modified glass slide. In Fig. 3a, the nitrogen 1s peak observed at around 400 eV indicates presence of APTES on the glass substrate. This peak is absent in the baseline XPS spectra of a glass slide without APTES termination. Fig. 3 also shows XPS spectra of APTES-terminated glass substrate after ink deposition and drying (Fig. 3b), ink sintering (Fig.  3c) and reduction (Fig. 3d), respectively. Titanium 2p 1 and 2p 3 peaks, observed at 464 eV and 458 eV respectively, confirm the presence of titanium on the surface after ink drying, sintering and reduction steps. Presence of palladium too is confirmed by the Pd 3d 3/2 and 3d 5/2 peaks observed around 341 eV and 336 eV, respectively.
After the initial drying step (Fig. 3b), strong chlorine 2s peak is observed at 269 eV indicating that Pd is present as its chloride in the dried ink. Upon sintering at elevated temperature (Fig. 3c), the intensity of the chlorine peak is significantly reduced. After ink reduction (Fig. 3d), the chlorine peak is absent. Further insights into the transitions occurring during the drying, sintering and reduction steps can be gained by observing the Pd 3d 5/2 XPS spectra (Fig. 4). Figs. 4a, 4b and 4c show the XPS spectra after drying, sintering, and reduction steps, respectively. Fig. 4a shows the highest binding energy at 338.0 eV. This binding energy, 11 together with the chlorine peak observed in Fig. 3b, suggests the presence of PdCl 2 in the dried deposit. After sintering, the palladium 3d 5/2 peak in Fig. 4b shifts to 336 eV, consistent with the reported binding energy for PdO. 12 This binding energy shift, along with the attenuation of the Cl 2s peak (Fig. 3c), suggests that the PdCl 2 from the dried ink transitions mostly to PdO during high temperature sintering. When the sintered PdO-TiO 2 is further reduced by the chemical reductant DMAB, the palladium 3d 5/2 peak shifts to approximately 335 eV (Fig. 4c). This shift indicates reduction of PdO to metallic Pd 12,13 embedded in a TiO 2 matrix. This Pd-TiO 2 interfacial layer provides adhesion enhancement as shown below.
It is worthwhile to note that a weak nitrogen 1s peak is still observed after PdCl 2 -TiO 2 ink deposition and drying (Fig. 3b), which indicates the presence of some APTES molecules on the glass surface. This peak is not evident after sintering (Fig. 3c) suggesting perhaps that the APTES monolayer is volatilized or decomposed at the higher sintering temperature. 14 Surface characteristics of Pd-TiO 2 deposits.-To characterize the surface during Pd-TiO 2 deposition, we collected surface profilometer scans at the various stages of deposition, i.e., for the untreated glass substrate (Fig. 5a), after drying the PdCl 2 -TiO 2 ink (Fig. 5b), after sintering to form PdO-TiO 2 (Fig. 5c), and finally after reduction to form the Pd-TiO 2 adhesion layer (Fig. 5d). Profilometer scans indicate that the underlying glass substrate is fairly smooth with RMS roughness of ∼7 nm. However, after ink application, sintering and reduction, sub-micron sized roughness elements are detected on the glass substrate as highlighted by arrows in Figs. 5b-5d. To observe the surface structure after Pd-TiO 2 formation, we used high-resolution optical microscopy (Fig. 5e), which confirmed the presence of sub-micron sized particles uniformly distributed on the glass substrate. While high-resolution SEM imaging and EDS compositional mapping were attempted, reliable results could not be obtained because of 'charging' on account of the poorly conducting glass surface. Nonetheless, XPS analysis (reported above - Figs. 3 and 4) provided valuable compositional information. The presence of Pd-TiO 2 particles provides anchoring sites that enhance interfacial adhesion while catalyzing subsequent electroless deposition as discussed below. We believe that the Pd-TiO 2 particle density may be modulated via the ink application, drying and sintering process conditions; however, further studies are required to understand how the particle density can be precisely controlled.
Adhesion characterization of Cu films.-The Pd-TiO 2 adhesionpromoting layers formed above served as catalysts for electroless seed-layer deposition followed by electrodeposition of Cu. Electroless Cu seed layers were favored over other metals such as Ni and Co because of the high electrical conductivity of Cu, which is essential for promoting uniform current distribution during the subsequent Cu electrodeposition step. Fig. 6 shows results of tape tests for Cu films deposited on a glass substrate modified with SAM and Pd-TiO 2 following the process outlined above (see Fig. 1 schematic). Tape tests were conducted using the procedure described above. Figs. 6a and 6b are images after tape testing of an electroless Cu plated glass substrate and the tape fragment, respectively. In this case, the ∼440 nm thick electroless Cu film remained attached to the substrate without interfacial failure. Figs. 6c and 6d are images after tape testing of a glass substrate modified with SAM and Pd-TiO 2 after electroless Cu seeding and thick (∼100 μm) Cu electroplating. Additive-free Cu electroplating baths were employed (which contained ppm-levels of chloride) because these baths are known to minimize stress in electroplated films. 9 Additives-containing electrolytes may be used after optimization of the plating conditions that provide minimal stress in thick deposits. The tape test results, even after aggressive cross-hatching, confirm the superior adhesion of thick (∼100 μm) electroplated Cu films to glass. The above results demonstrate the adhesion enhance- ment provided by interfacial SAM/Pd-TiO 2 adhesion-promoting layers in enabling Cu metallization of glass substrates. In the absence of the SAM layer, reliable Pd-TiO 2 adhesion layers could not be formed and the resulting Cu films peeled off either during electrodeposition or during subsequent tape testing. It is also worthwhile to indicate that SAM/Pd layers fabricated without the TiO 2 matrix layer provided reasonable interfacial adhesion enhancement for thin (∼100 nm) Cu layers; however, could not enable thick (>10 μm) Cu coatings.
In order to quantify the adhesion of electrochemically deposited Cu on glass using the sequence outlined in Fig. 1   with ∼400 nm electroless Cu seed-layer and 15 μm electroplated Cu. Fig. 7 shows results of a 90 • peel strength test. The average observed peel strength was 1.4 N/cm and the maximum peel strength was 1.8 N/cm. These peel strength values are comparable to those reported by Shen and Dow for adherent Cu films on AlN substrates enabled using chemical grafting. 15 While the above adhesion testing establishes the basic feasibility of an 'all-wet' route for depositing thick, adherent Cu films on glass substrates, further process optimization and characterization is needed to implement the process on large-area substrates and on patterned surfaces.

Conclusions
An 'all-wet' process for Cu metallization of glass substrates has been demonstrated. In this process, interfacial adhesion is promoted through the use of SAM/Pd-TiO 2 adhesion layer that also catalyzes electroless deposition of a Cu seed layer. The mechanism for Pd-TiO 2 adhesion layer formation involves stepwise reduction of Pd +2 to metallic Pd embedded in a TiO 2 matrix. Electrodeposition of Cu onto the electroless Cu seed layer enables thick (>10 μm), adherent Cu films. The proposed 'all-wet' strategy provides a significant adhesion enhancement as observed in tape tests and 90 • peel strength tests. Owing to it's low-cost and ease of integration, the 'all-wet' process offers promise in numerous metallization applications where insulator-metal interfaces are encountered.