Direct observation of nanowire growth and decomposition

Fundamental concepts of the crystal formation suggest that the growth and decomposition are determined by simultaneous embedding and removal of the atoms. Apparently, by changing the crystal formation conditions one can switch the regimes from the growth to decomposition. To the best of our knowledge, so far this has been only postulated, but never observed at the atomic level. By means of in situ environmental transmission electron microscopy we monitored and examined the atomic layer transformation at the conditions of the crystal growth and its decomposition using CuO nanowires selected as a model object. The atomic layer growth/decomposition was studied by varying an O2 partial pressure. Three distinct regimes of the atomic layer evolution were experimentally observed: growth, transition and decomposition. The transition regime, at which atomic layer growth/decomposition switch takes place, is characterised by random nucleation of the atomic layers on the growing {111} surface. The decomposition starts on the side of the nanowire by removing the atomic layers without altering the overall crystal structure, which besides the fundamental importance offers new possibilities for the nanowire manipulation. Understanding of the crystal growth kinetics and nucleation at the atomic level is essential for the precise control of 1D crystal formation.

first clear and atomically resolved transition between the growth and decomposition on the surface of NWs at the AL. We observed a clear AL nucleation character change during the transition from the growth to the decomposition of CuO NWs. Interestingly the layer-by-layer decomposition of NWs happens without disrupting its crystal structure. This study reveals the kinetics and mechanism of the AL arrangement during the crystal growth at oxidative conditions, the decomposition in vacuum and the switch between these two processes. The understanding of the crystal transformations at the AL during the NW growth is crucial to control the NW structure, composition and thereby its properties. The investigations of the CuO NW growth and decomposition are important from the fundamental point of view, as the understanding of the 1D crystal formation mechanism can be extended to other systems.

Results and Discussion
Initially, CuO NW growth was ex situ observed by heating Cu samples at the pressure ranging from high vacuum to the atmospheric pressure (Fig. 1). It was found that the NW growth occurs at O 2 pressures above 400 Pa, and the density of the grown NWs increases along with the pressure. At pressures below 400 Pa, the NW growth was not observed. In the pressure range from 20 Pa to 400 Pa, faceted crystals were formed on the surface (Supplementary Fig. S1). At pressures below 20 Pa no growth of 1D structures was observed, only a planar layer of CuO grew.
The in situ environmental transmission electron microscopy (ETEM) observations of CuO NWs under changing O 2 pressure at a constant temperature (400 °C) revealed 3 AL evolution regimes: the growth, transition and decomposition. First, as it was observed in ex-situ experiments, the formation of NW structure, caused by ordered AL growth, was observed only over 400 Pa. The CuO NW growth followed layer by layer manner, when every subsequent AL grows exactly on top of a primary one ( Supplementary Fig. S2). ALs nucleated at the edge of the twin boundary ridge, at the tip of NW ( Fig. 2A). At the initial O 2 pressure under 400 Pa, ordered AL growth (NW growth) was not observed, ALs were nucleating randomly and a planar layer of CuO was grown. Interestingly, if the NW formation started at higher pressures (in the growth regime) and the pressure subsequently was reduced during the observation, the ordered growth of ALs proceeded even at lower pressures (from 100 to 400 Pa), even though that ALs nucleate randomly on the growing surface. Figure 2 shows the results obtained during the in situ observation of the NW growth initiated at 700 Pa and continued after the O 2 pressure was reduced to 340 Pa (Supplementary Video S1). The video demonstrates that O 2 pressure is critical for the initial ordered nucleation of ALs (NW formation), but if the NW structure is already available, ALs nucleate at the growing surface, retaining the ordered structure.
During the formation of ALs in the growth regime, ALs nucleate at the twin boundary of the growing NW ( Fig. 2A), however, in the transition regime the AL nucleation is observed at random points on {111} surface (Figs 2B, 3). This shows that the nucleation conditions changed and the twin boundary is not anymore the preferential nucleation point. Interestingly, the AL growth rate 1.7±0.4 nm/s in the transition regime remained in the same order of magnitude as in the growth regime (1.4±0.3 nm/s) even if the O 2 pressure was lower (Fig. 4). The FFT pattern (Fig. 3A inset) can be well indexed using the monoclinic CuO structure (a = 4.69, b = 3.43, с = 5.13 Å, β = 99.55°), which is consistent with our earlier analysis [15][16][17] .
Further decreasing the O 2 pressure in the system led to the decomposition regime. Experimentally it was found that the decomposition regime occurred at the O 2 pressures around 0.4 Pa. It can be noticed that the decomposition of NWs takes place in an ordered layer-by-layer manner, similarly to the ordered growth of ALs on the NW tip, but with longer induction time and higher AL decomposition rate of 4.2±0.5 nm/s (Fig. 5). Induction time is a period of time that is necessary to initiate a layer growth, and it was calculated on the basis of ETEM images as the time between nucleation of two consequent ALs.
The decomposition started from the base of the CuO NW on (002) lattice plane. We did not observe the transition in the crystal phase to any intermediates such as Cu 2 O or to metallic Cu. In case of CuO reduction in vacuum, considerable changes can be seen already at 100°C 13 and the most probable mechanism involves oxygen diffusion to bulk Cu rather than oxygen desorption into vacuum 13,14 . In our case, the decomposition of ALs starts at the base of the NW, which may be related with the shorter diffusion route from the base of NW to the bulk Cu, located underneath the surface (Supplementary Video 2, Supplementary Figure S3).  On the basis of the experimentally determined rate of the AL decomposition (v dec = 4.2 nm s −1 , Fig. 5), we can calculate the decomposition time of CuO as l/v dec , where l = 0.27 nm is the size of CuO molecule, which can be estimated from the bulk density and volume per one molecule (0.021 nm 3 ). Thus, we obtain the experimental decomposition time of τ = 0.064 s.
We can also estimate the CuO decomposition time based on the decomposition energy, E, and the frequency of the CuO lattice vibration, ν, as 3 :   respectively. These values are in a good agreement with the decomposition time determined experimentally (0.064 s). Bearing in mind that during the CuO AL formation two processes -embedding of CuO to the layer and removal of CuO from the layer -are conducted simultaneously. Therefore, let us compare the arrival time of "active" O 2 molecules (forming CuO), and the removal time of CuO from the layer, defined from eq. (1). The arrival time of "active" O 2 molecules to the CuO molecule can be estimated similar to Eq. 1 as: a g a 1 where ν g = l 2 Vn/4 is a frequency of gas molecule collisions with wall of area l 2 , V = (8RT 0 /πM ox ) 0.5 is a mean velocity of O 2 molecules, n = pN A /RT 0 is a number concentration of O 2 inside ETEM chamber, p is an O 2 pressure, and M ox is an O 2 molar weight, N A is the Avogadro number, T 0 is a temperature of oxygen inside the ETEM chamber, E a is the activation energy of NW growth. Substituting E a = 37 kJmol −1 , T = 673 K, and T 0 = 298 K 17 and the experimental deposition time (τ a = 0.064 s) to Eq. (2) we obtain p = 6 Pa. This means that at the O 2 pressure lower than 6 Pa the removal from the AL prevails, which explains the experimentally observed decomposition at 0.4 Pa. At the same time, the AL growth is observed at 400 Pa, the pressure which is almost 70 times higher than the pressure above which the CuO formation is preferred. We believe that the excess pressure is related to the formation of the NW, since as we demonstrated above, if the NW structure is already formed, the NW growth proceeds at lower pressure.
It is worth noting that the AL nucleation is usually more favorable at the corner of the facet of the growing crystal, where the supersaturation is higher than at the center (the Berg's effect) 19,20 . However, as we found from the in situ observation, with the pressure decreasing the AL growth may start at any surface point. The nucleation of the AL at random place may be determined by a weakening of the Berg's effect due to the decrease of CuO molecule concentration gradient along the edge at lower O 2 pressure.
Thus, the O 2 pressure affects both interconnected processes: the growth and decomposition. Obviously, there must be such O 2 pressures at which the 1D growth ceases and at which decomposition prevails. We experimentally found these O 2 pressures to be around 400 Pa for the AL growth and 0.4 Pa for the decomposition, respectively.

Conclusions
To summarize, an atomistic examination of the CuO NW evolution in the conditions ranging from oxidative to reductive ones was carried out. The growth, transition and decomposition regimes were experimentally detected and explained from the AL kinetics point of view. The AL growth/decomposition regimes are determined by two simultaneous processes: embedding and removal of CuO from the AL, where both processes are in equilibrium at the O 2 pressure of 6 Pa. Therefore, the experimentally observed AL decomposition was obtained at pressures below 0.4 Pa as removal process prevails at lower O 2 pressures. The NW growth is observed at pressures above 400 Pa. The range between those pressures associated with the transition regime, where crystal growth conditions are favorable only in case of previously formed NWs. Moreover, the transition regime is characterized by weakening of the Berg's effect, which leads to the change of the AL nucleation from the edge of the twin boundary ridge at the NW tip to a random position on the surface at lattice plane {111}.
Decomposition of CuO NW proceeded in ordered layer-by-layer manner at 0.4 Pa from the bottom side of the NW at the lattice plane (002), which can be explained by the shorter diffusion route from the NW base to underlying bulk Cu. A crystal phase transformation from CuO to intermediate Cu 2 O or Cu was not observed.
Although we investigated the AL growth and decomposition only in one kind of NWs, these atomistic principles can be common to the other 1D crystals. Since the decomposition was pronounced in the bottom part of the NW, it can be also used as a method for NW diameter tailoring or removal from the substrate.

Methods
An aberration-corrected FEI Titan 80-300 ETEM operated at 300 kV was employed for the in situ observation of the AL evolution 17,21 . The sample was dispersed in ethanol and drop-casted on a MEMS-based micro-heater from DENSsolutions. After drying at room temperature, the sample was inserted into the electron microscope.
The CuO NWs were synthesized in situ in the ETEM by simple oxidation of pure metallic Cu (>99.99%) powder in the presence of O 2 . The decomposition of the NWs was carried out by lowering the O 2 pressure maintaining the same temperature of the substrate. The NWs were grown directly on the as-received Cu powder, with no patterning or catalyst addition. The observation time at the set temperature was 30-200 min. Typically, the CuO NWs were observed at 400 °C varying the O 2 partial pressure from 10 -6 up to 700 Pa. The electron beam dose varied depending on the conditions from 9 × 10 3 to 9 × 10 4 electrons/(nm 2 s). No visible electron beam damage was observed during the in situ growth of the NWs, also no visible difference between the observation area and the rest of the sample was found. For the ex situ growth of NWs, pure metallic Cu (99.999%) samples, were heated for 30 min in a vacuum tube furnace (inner diameter 50 mm) Carbolite CTF 12/65/550 with built in flowmeter for O 2 and control unit 3216P1 at O 2 partial pressures from 20 Pa to the atmospheric pressure. After the synthesis, the samples were transferred directly to scanning electron microscope (SEM, JEOL JSM 7500F) as grown, without any pre-treatment.