**1. Introduction**

Copper is ubiquitous in numerous industries due to its relatively low cost coupled with its malleability and high electrical and thermal conductivities. It has uses in everything from residential and commercial plumbing [1–4] and electrical wiring to industrial heat exchangers [5–7] and high-powered electronics [8–11]. Copper possesses adequate aqueous corrosion resistance due to the formation of a semi-protective native oxide film. Still, it corrodes at a finite rate and is susceptible to pitting dependent on solution constituents, pH, and temperature [2–4,12–16].

One specific application of interest here is the use of copper in radio frequency (RF) devices (i.e., traveling-wave tubes and crossed-field amplifiers). These devices typically contain fluid-cooled copper collectors and/or copper cooling channels [17–24]. Often, the cooling fluid is specified as deionized (DI) water with low dissolved oxygen content, in which case corrosion of the wetted copper components is insignificant within the useful lifetime of the device. However, the presence of contaminants—such as chlorides, additional oxygen, and carbon dioxide—in the cooling water supply can accelerate the corrosion to unacceptable levels. Chlorides are known to be particularly aggressive toward copper [12,14,25], as with many other metals, and are common contaminants in water supplies. Copper readily forms salt compounds in the presence of chlorides and can have its protective oxide film locally disrupted, allowing for pitting to occur.

The corrosion of copper in chloride media has been thoroughly studied [13–15,26–30], from which it has been concluded that the anodic and cathodic half reactions are given by Equations (1) and (2), respectively. The anodic reaction proceeds in two steps with the mostly insoluble CuCl produced in the first step and soluble CuCl2− produced in the second. CuCl builds up as a film on the Cu surface, leading to passivation, or, more appropriately, pseudo-passivation. Higher chloride concentrations tend to shift the equilibrium of Equation (1) toward CuCl2− and can even produce higher chloride complexes (i.e., CuCl3<sup>2</sup>− and CuCl4<sup>3</sup>−) [27,31]. This increases dissolution of the CuCl surface film and accounts for the more aggressive corrosion seen in solutions with higher chloride content. Additionally, cuprous oxide (Cu2O) may be produced from CuCl2<sup>−</sup>, but its stability also decreases as the chloride content increases [31].

Oxygen reduction and water reduction (Equation (2)) dominate the cathodic current, though there are complications in the presence of copper corrosion products and surface films [31], including the reduction of CuCl on the copper surface [30]. Water reduction dominates when dissolved oxygen levels are low, whereas oxygen reduction produces high cathodic currents in oxygenated solutions [32]. Although the corrosion of copper in pure water via water reduction and hydrogen evolution in the absence of oxygen is still under debate [33–37], it is reasonable to expect that this reaction could play a role in the presence of chlorides and at sufficient overpotential.

$$\begin{array}{l}\text{Cu} + \text{Cl}^- \rightarrow \text{CuCl} + \text{e}^-\\\text{CuCl} + \text{Cl}^- \rightarrow \text{CuCl}\_2^- \end{array} \tag{1}$$

$$\begin{aligned} \text{O}\_2 + 2\text{H}\_2\text{O} + 4\text{e}^- &\rightarrow 4\text{OH}^- \\ 2\text{H}\_2\text{O} + 2\text{e}^- &\rightarrow \text{H}\_2 + 2\text{OH}^- \end{aligned} \tag{2}$$

A common approach to improve the corrosion resistance of copper is to alloy it with other metals, such as aluminum and nickel. However, alloying is not always an option, as it can significantly reduce the electrical (and thermal) conductivity [38]. Another option is the application of barrier coatings to the copper surface. To preserve the desirable bulk properties of the copper, the barrier coatings should be as thin as possible. Additionally, copper components of RF devices are often of complex geometry or high aspect ratio, which narrows the available thin film deposition techniques.

Atomic layer deposition (ALD) is a vapor-phase technique that is uniquely suited for deposition of conformal thin films over complex, non-planar surfaces and through high-aspect-ratio structures, like tubing. ALD has no line of sight requirement with sub-nanometer thickness control [39–43]. ALD can also proceed at low temperatures (<150 ◦C) [44,45], avoiding temperature limits of fabricated components and even enabling deposition on thermally-sensitive substrates, such as polymers [46–50] and microelectronics [51,52].

Alumina (Al2O3) and titania (TiO2) are two of the most widely-studied ALD processes. Deposition of alumina via trimethylaluminum (TMA, Al(CH3)3) and H2O is thermodynamically favorable and nucleates well on most surfaces [39,40,50,53]. Al2O3 has excellent sealing properties and has shown outstanding versatility as a barrier layer on metals [54–58], polymers [48,59–62], and electronics [63–67]. However, the use of Al2O3 as a corrosion barrier is limited by its chemical stability and dissolution in alkaline media [68,69]. On the other hand, titania is lauded for its chemical stability [68,70–72] and has shown favorable corrosion resistance [73–78]. Titania ALD, however, exhibits film nucleation issues using titanium tetrachloride (TiCl4) as a precursor, especially on copper substrates, and tends to deposit with high roughness, leading to increased porosity and poor corrosion performance over time [73,75,79].

Previous studies of ALD films for corrosion protection of copper have mostly focused on aluminum oxide. Mirhashemihaghighi et al. [80] reported a 7x increase in polarization resistance over uncoated, polished copper with a 10 nm ALD Al2O3 film and a three order of magnitude increase with a 50 nm Al2O3 film in deoxygenated 0.5 M NaCl. Chai et al. [81,82] found better corrosion protection with increasing alumina film thickness in aerated 0.1 M NaCl, as did Daubert et al. [79]. Daubert et al. [79]

also performed electrochemical impedance spectroscopy (EIS) over 90 h for 5 different metal oxide ALD films, showing a decrease in high-frequency impedance for the Al2O3 film and better stability for the other four. A study by Abdulagatov et al. utilized an alumina base layer and titania capping layer to overcome the limitations of the individual layers, resulting in significantly improved corrosion protection of copper over 900 h in water at 25◦ and 90 ◦C [73]. Although not specifically related to aqueous corrosion, it is worth noting that Chang et al. successfully applied 100 nm thick ALD alumina films to copper for protection against oxidation in air at 200 ◦C [54]. Whereas ALD alumina possesses excellent sealing properties and can provide adequate initial corrosion protection with films less than 50 nm thick, its longer-term stability on copper is in question. Preliminary results of combining alumina and titania prove promising as a corrosion barrier, and further investigation into nanolaminate alumina/titania film structures is warranted. Reports of Al2O3/TiO2 nanolaminated ALD thin films for corrosion protection of steel are available [75,83,84], showing increased corrosion and delamination resistance over single-layer films.

Here we investigate the corrosion behavior of ALD-coated copper and use Al2O3/TiO2 nanolaminate films to enhance corrosion protection over the single-layer materials. The effect of dissolved oxygen on the corrosion protection of these ALD thin films has previously not been reported. Oxygen is always present in some amount, even when low oxygen levels are maintained in contact with the copper surfaces. Determining its impact on the protective performance of ALD films is essential for their use in RF devices and other copper-containing components. Also of interest is the effect of ALD film layer structure, copper-based interfacial films, and deposition temperature on the corrosion behavior of ALD alumina- and titania-coated copper. In this work, we utilize DC voltammetry and electrochemical impedance spectroscopy (EIS) to investigate corrosion behavior and probe the stability of Al2O3/TiO2 ALD films on copper in a sodium chloride (NaCl) solution with and without dissolved oxygen.

#### **2. Materials and Methods**

#### *2.1. Material Preparation*

#### 2.1.1. Substrate Preparation

Grade 110 copper sheets (99.9%, McMaster-Carr, Douglasville, GA, USA) were cut into 15 mm × 25 mm × 1 mm coupons. Coupons had a No. 8 mirror finish with average surface roughness of 0.1–0.3 μm. Prior to loading into the ALD reactor, all coupons were ultrasonically rinsed in acetone for 5 min, rinsed with isopropanol and DI water, then dipped in 35% H3PO4 for 30 s to reduce the native copper oxide, and finally rinsed in DI water and dried thoroughly with ultra-high-purity nitrogen.

#### 2.1.2. Thin Film Deposition and Characterization

Films were deposited in a home-built, viscous-flow, hot-wall ALD reactor using ultra-high-purity nitrogen (N2—99.999%, Arc3 Gases, Richmond, VA, USA) as the carrier and purge gas. Nitrogen was constantly flowing during deposition, maintaining a reactor pressure of roughly 1 torr. Alumina (Al2O3) was deposited using trimethylaluminum (TMA, 98% Strem Chemicals, Newburyport, MA, USA), and titania (TiO2) was deposited using titanium tetrachloride (TiCl4, 99% Strem Chemicals, Newburyport, MA, USA). DI water served as the co-reactant for both precursors. Each ALD cycle consisted of a 0.1 s precursor (TMA or TiCl4) dose followed by a 45 second N2 purge and a 0.1 s H2O dose followed by a 45 s N2 purge. Unless otherwise specified, Al2O3 and TiO2 were deposited at 150 ◦C with nominal growth rates of 1.25 Å/cycle and 0.45 Å/cycle, respectively, measured using spectroscopic ellipsometry (J.A. Wollam Co., Lincoln, NE, USA) on the copper coupons and on silicon monitor wafers coated simultaneously with the copper coupons. A Keyence VKx1100 confocal laser scanning microscope (Itasca, IL, USA) equipped with a 404 nm violet laser was used for imaging, and

energy dispersive X-ray spectroscopy (EDX) was performed using a JEOL 6010LA scanning electron microscope (Peabody, MA, USA, results not shown).
