Oxidation at low temperature (200–350 ˚C)
Under low temperatures (200–350 ˚C), small (< 10 nm) oxide islands were observed to form on the faceted holes. The oxide islands are similar to native oxides on Cu films formed at room temperature. Using in situ ETEM, the nucleation sequence of nano-islands on different facets is characterized. Due to the low temperature, the oxidation speed is much lower at the same oxidizing pressure. For example, at 250 ˚C under 0.03 Pa O2, it took two hours of oxygen exposure until oxide nano-islands were observed. To accelerate the oxidation process, higher oxygen pressure (0.1 Pa) was used at low temperatures. This reduced the amount of time needed to form observable oxides to tens of minutes.
As shown in Fig. 1, oxide islands were first observed on Cu(110) facets at 250˚C under 0.1 Pa O2 (Fig. 1a-b), then they were seen on Cu(100) facets (Fig. 1c). The time sequence over which each oxide island formed is plotted in Fig. 1d, which shows a clear trend of oxide nucleation preference on Cu(110) facets over Cu(100) facets. During this process depicted in Fig. 1e-f, oxide nano-islands are mainly observed along the hole, rather than on the flat surface. They mostly demonstrate outward oxidation behavior, that is, oxide growth on oxide surfaces without significantly changing the underlying metal surfaces. These results indicate diffusing Cu atoms near the holes are the major Cu source for oxides during this stage. This is consistent with our previous HRTEM observation of oxide growth at a similar condition (300 ˚C, 0.03 Pa O2)[16], which shows layer-by-layer oxide growth on the Cu2O surface. Such growth occurs without consumption of the metal substrate under the oxide island, until surface diffusing Cu atoms are depleted.
During the early oxidation, oxide nano-islands were only observed near the holes (Fig. 1e-f). After a longer oxidation time (Fig. 1g), oxide nano-islands were observed on Cu(001) thin film surfaces. Selected area electron diffraction (SAED) patterns (Fig. 1h) show Cu2O nano-islands follow cube-on-cube epitaxy with Cu(001) substrates. Oxide islands on Cu surfaces demonstrate homogenous, uniform distributions along surfaces, similar to previous experimental observations on flat Cu surfaces [17,18,21]. Over this temperature range, an oxidation preference sequence of Cu(110) facet > Cu(100) facet > > Cu(001) flat surface is resolved. Thus, holes can significantly accelerate oxidation by introducing edges, preferable facet orientations, and free Cu atoms diffusing on surfaces.
Oxidation at medium temperature (350–450 ˚C)
Under a medium temperature range (350–450 ˚C), larger oxides (~ 100 nm in at least one dimension) are observed to form along Cu(110) facets (Fig. 2). As shown in Fig. 2a, needle-shaped oxides were observed to nucleate on Cu(110) facets at 350 ˚C. As depicted by the oxide on the rightward (110) facet, nucleating Cu2O initially occurs only on {100} facets, growing in both width and height while maintaining its triangular shape. When the height of the oxide reached ~ 15 nm, a {110} interface formed at its bottom. This resulted in a trapezoid-shaped oxide with {100} side facets and {110} bottom facets (Fig. 2b). Then, this oxide grew mainly along its width, while its height remained almost constant. Both high-resolution (HRTEM) image (Fig. 2b) and the SAED pattern (Fig. 2c) suggest Cu2O islands formed under this condition also share cube-on-cube epitaxy with the Cu substrate.
In contrast, few oxides were observed to nucleate on Cu(100) facets. Oxides nucleating on Cu(100) facets are observed only on larger holes (Cu(100) facet > 100 nm), not on smaller holes. This is probably due to the fact that on smaller holes the distances between oxides -- from oxides that first nucleate on Cu(110) and oxides that will subsequently nucleate on the Cu(100) facets -- are insufficient for the average oxide distance under this condition. Only larger holes with large Cu(100) facets can provide enough space for the widely spaced nucleating oxides. On the Cu(001) flat surface, right-triangle-shaped Cu2O are observed (Fig. 2d). They feature {100} short edges and {110} long edges, respectively marked by dashed red and blue lines. These triangular-shaped oxide islands are consistent with previous literature reports on uniform Cu thin films.[17,30]
At 450˚C, oxides are also observed to nucleate on Cu(110) facets first (Fig. 2e), forming rectangular-shaped oxides with {110} facets. On flat Cu(001) surfaces, right-triangular-shaped Cu2O islands are also observed, exposing {100} short edges and {110} long edges (Fig. 2f). Compared with outcomes at 350 ˚C, oxide densities decrease and average oxide distances increase, while larger oxide islands were observed after the same oxidation time.
Under this temperature range, oxide formation follows the sequence of Cu{110} facet > Cu{100} facet ≥ Cu(001) surface. Oxides grown on facets show both inward and outward oxidation, that is, oxides grow both on the metal-oxide interface and on the oxide surface. These results indicate that under these conditions, Cu atoms from facets are the main Cu source for Cu2O growth on holes. During the oxidation process, O from faceted holes serves as the main O source for Cu2O, such that O needs to diffuse through the oxide to form new Cu2O. This leads to self-limiting oxidation in the depth direction < 110>, which forms observed needle-shaped islands on Cu(110) facets.
Oxidation at high temperatures (500–600˚C)
Under high temperatures (500–600 ˚C), oxide islands form much faster and grow into larger and more faceted shapes. Unlike lower temperatures that prefer nucleation at Cu(110) facets, oxides at high temperatures prefer to nucleate at Cu(100) facets (Fig. 3).
At 500 ˚C, a mixture of oxides nucleating on both Cu(100) and Cu(110) facets are observed, indicating that their nucleation preferences are similar (Fig. 3a). Oxides nucleating on Cu(110) facets feature {100} (red) and {110} (blue) interfaces, similar to those formed in the medium-temperature range. However, the oxides grown on Cu(100) facets show a directional growth into Cu substrates along the < 110 > direction, forming rods. The sides of the rods are oriented along {024} facets (green).
At 550 ˚C and 600 ˚C, oxides are only observed to nucleate on Cu(100) facets (Fig. 3b-e, Fig. 5c). These oxides feature higher index facets like {024} (green), and lower index facets such as {100} (red) and {110} (blue). As shown in Fig. 3c-e, the metal-oxide interfaces along the < 100 > and < 110 > directions grow much faster than along the < 024 > directions. This is evidenced by the size evolution of the oxide, and the consumption of Cu near the Cu2O||Cu{100} interface, as the originally faceted hole (dashed line) became concave.
As shown in Fig. 3f, oxide nucleation on Cu(100) facets requires a relatively large Cu(100) facet. When the Cu(100) facet is too small, Cu2O is observed to form only on Cu(001) surfaces, forming square-shaped islands exposing {110} facets. This oxide shape on flat surfaces is consistent with previous literature reports at this temperature.[17] Oxides forming on flat surfaces show random distributions, regardless of their relative distances from holes. This result indicates oxide nucleation on Cu(001) surfaces is preferred over that on Cu(110) facets at high temperatures.
After prolonged oxidation, all holes with Cu(100) facets are oxidized, forming zig-zag shaped oxide rods bending at holes (Fig. 4a-c). The rods show side facets along {013} (yellow), {024} (green), or {015} (purple) orientations, and a top facet along {010} (red). As shown in Fig. 4, the oxide grows mainly in the length direction (along < 010>), while the width of the rods maintains almost unchanged. This is similar to the observation in Fig. 3b-d, indicating the Cu2O||Cu < 010 > direction is the preferred oxide growth direction. Figure 4d plots each actual oxide rod length between {020} rod tip and the nearest hole facet, while Fig. 4e measures the relative length changes from Fig. 4(a). Both lengths are measured along the < 010 > direction perpendicular to the rod tip surface ({020}, as illustrated in Fig. 4c. Figure 4e confirms the growth rates of each rod are similar (3.8 ± 0.5 nm/min), following a quasi-linear relationship with time regardless of the initial length (Fig. 4d), width, or side facet orientation of each rod. This result indicates that the O source for Cu2O rod growth is coming from absorbed O on Cu film surfaces, instead of O on holes that needs to diffuse through the oxide. Because if the latter case was true, growth rate would decrease with increasing oxide thickness.
Hence, in this high-temperature range (500–600 ˚C), the oxidation preference sequence follows Cu(100) facet > Cu(001) surface > > Cu(110) facet. The oxidation of Cu(100) faceted holes demonstrates inward oxidation, that is, oxide growth into metal, without expanding on the original metal surface. O on the Cu surface serves as the O source for Cu2O growth rather than O from the faceted hole, which was observed at medium and low temperatures.