Experimental subsection

The first of the seven seed layers was the Zn terminated surface of a c-axis oriented, hydrothermal ZnO single crystal (CrysTec GmbH; denoted as “BULK”). The second ZnO seed layer was an epitaxial ZnO thin film grown by pulsed laser deposition (PLD) on c-sapphire (denoted as “PLD/c-sapphire”) using a KrF excimer laser (248 nm) [101]. The third ZnO seed layer was ZnO grown by atomic layer deposition (ALD) onto a thick (~6µm) epitaxial MOCVD GaN film (TDI Corporation) on c-sapphire (denoted as “ALD/GaN”). The layer was grown with 500 ALD cycles at 300 ºC deposition temperature [102]. The fourth ZnO seed layer was ZnO grown by PLD on Si(111) (denoted as “PLD/Si”) [103]. The fifth ZnO seed layer was also deposited by PLD onto Si(100), but this time a 130 nm thick thermal SiO2 was grown prior to the ZnO growth. This sample was denoted as “PLD/SiO2/Si” [103].

The sixth ZnO seed layer was again a PLD ZnO on c-sapphire, but here a 100 nm thick highly

53

textured (111) oriented Pt buffer layer was deposited before the ZnO growth in a DC magnetron sputtering system, as described elsewhere [104]. This sample was denoted as

“PLD/Pt/sapphire”. The last seed layer was grown on a Si(100) wafer in Ar/O2 atmosphere by DC reactive magnetron sputtering of an Al doped (2 wt %) Zn target (denoted as “Sputt/Si”).

In order to stabilize the glow discharge, a pulse signal was added to the direct current voltage.

The deposition was carried out without bias voltage or substrate preheating [105]. It should be noted that no attempt was made to remove the native amorphous SiO2 passivation layer on the surface of either the (111) or (100) Si substrates. PLD/sapphire, PLD/SiO2/Si and PLD/Pt/sapphire were deposited at the National Institute for Materials Science (NIMS) in Japan, ALD/GaN and Sputt/Si were deposited at MTA TTK MFA, and PLD/Si seed layer was made at Nanovation SARL in France.

The process flow for the fabrication of the ZnO NWs is shown in Fig. 19 [90, 93]. First of all each seed surface was washed ultrasonically in acetone, ethanol, and deionized water (Fig.

19a). In order to provide patterned templates for subsequent NW growth, a ∼300 nm thick poly(methyl methacrylate) (PMMA) resist layer was spin coated onto each seed surface, and a regular triangular lattice (150 μm x 150 μm, with a lattice constant of 500 nm) of circular growth windows (∼120-130 nm in diameter) was then exposed in the PMMA layer by e-beam lithography (Fig. 19b). After development, NW arrays were synthesized using a low temperature wet chemical method (Fig. 19c) based on an aqueous solution having the same concentrations (4 mM) of zinc nitrate hexahydrate (Zn(NO3)2·6H2O) and HMT ((CH2)6N4), as reported elsewhere [23]. During NW growth the solution and the specimens were in a tightly closed glass bottle (Fig. 19f), which was placed into a multipurpose oven. The specimens were mounted upside-down on a polytetrafluoroethylene (PTFE) sample holder in order to prevent any precipitates that formed in the nutrient solution from falling onto the substrates which would have inhibited the growth of the NWs. Fig. 19e shows the plot of the air temperature profile inside the oven during nanostructure growth as a function of time (black squares): at first the temperature was increased gradually by 60 °C/h up to 85 °C and was then held at 85 ºC for 2 hours. Then the heating was turned off and after an overnight cooling the sample was removed from the growth solution. Nevertheless, the actual temperature of the nutrient solution during nanostructure growth does not reach even 80 °C, as plotted by red circles in Fig. 19e. At solution temperature of 62 °C the nutrient solution became cloudy indicating the initiation of the reaction (dashed line in Fig. 19e). Finally the PMMA layer was

54

removed in acetone and the specimens were thoroughly rinsed with deionized water (Fig.

19d).

Figure 19. Schematic process flow of ZnO NR arrays. The fabrication steps are: surface treatment of ZnO seed surface (a), pattern generation in PMMA by e-beam lithography (b), chemical NW growth (c), and PMMA removal (d). Measured air temperature vs. time inside the multipurpose oven (black squares) and the corresponding actual solution temperature vs. time (red circles) during nanostructure growth (e). The dashed line indicates the initiation of the reaction. The NRs were synthesized in an aqueous solution: the specimen was mounted upside-down on a PTFE sample holder (f).

The resulting ZnO NW arrays were imaged by FE-SEM. Before nanostructure growth, the surface of the substrates was studied by AFM in semicontact mode. The thickness of the layers was determined by means of spectroscopic ellipsometry (Woollam M2000D rotating-compensator ellipsometer). The resistivity of the seed layers was measured using a four point probe method.

The crystal structure of BULK, PLD/sapphire, PLD/Si, PLD/Pt/sapphire and Sputt/Si seed layers and the corresponding NW arrays was investigated using high resolution XRD with a 50 μm diameter spot size, i.e. sufficiently smaller than the patterned areas. The XRD measurements were carried out at NIMS in Japan. A section of the Debye-Scherrer ring was studied using a 2D-detector system. In this approach, a 2θ/χ mapping around the (0002) peak is acquired simultaneously. The dispersion in the crystallographic alignment can then be estimated from the full width at half maximum (FWHM) along the χ direction while the

55

distribution of the c lattice parameter (and, hence, strain information) can be deduced from the FWHM along the 2θ direction [106].

In addition a cross-sectional TEM lamella representing the PLD/Pt/sapphire layer with the corresponding NWs was prepared and examined using TEM. The lamella was cut using FIB etching in the Zeiss 1540XB cross-beam system equipped with nanomanipulator. In order to protect the NWs from amorphization during the milling, photoresist was infiltrated at first into the NWs using spin coater. Hence the whole length of the NWs was covered with the protective polymer. The area to be thinned could be localized with micrometer accuracy on the sample surface by the high resolution FE-SEM and a high energy (30 keV, 5 nA) Ga⁺ ion beam was used for sputtering. The thinning process of FIB was monitored by a scanning electron beam (Fig. 20a). After cutting out, the lamella was lifted out by the nanomanipulator and fixed onto a TEM grid inside the SEM chamber using SEM glue (Fig. 20b). The final thinning was made on the TEM grid. In order to minimize the thickness of the amorphized region at the front- and backside of the lamella, the final thinning was carried out using lower current (10-100 pA) Ga⁺ beam.

Figure 20. SEM micrographs of the cross-sectional TEM lamella preparation. Protecting photoresist was infiltrated into the NWs, the area to be thinned was localized, and the thinning process of FIB was monitored by the electron beam (a). The lamella was lifted out by nanomanipulator and glued on the TEM grid in situ inside the SEM chamber (b).

The piezoelectric activity of individual NRs was characterized by piezoresponse force microscopy (PFM) where the mechanical response of the investigated NRs was recorded on an alternating voltage signal. Our SmartSPM multimode scanning probe microscope mounted with a conductive AFM tip was used at an excitation signal of frequency of 52 kHz and amplitude of 6V. In order to determine the d₃₃ constant (corresponding to longitudinal expansion/contraction along the c-axis in response to the applied electric field) of the inverse piezoelectric tensor represented by a 3×6 matrix in Voigt notation (see subsection 12.2. in

56

Appendix) for the NR the major part of the applied voltage has to be dropped on the NR, that is the resistance of the substrate is to be minimized. For this purpose NRs grown on PLD/Pt/sapphire were used, since a highly conductive Pt thin film of 100 nm exists beneath the ZnO seed layer of ~218 nm resulting in an overall sheet resistivity of 2.3 Ω/sq and providing an excellent bottom electrode for the NRs. In order to avoid the fracture of the NRs in contact mode scanning thicker NRs were synthesized for this particular purpose. However, in this case I applied a growth temperature of 95 °C (instead of 85 °C) and a concentration of 10 mM (instead of 4 mM), and wider nucleation windows (~500 nm) were generated in the PMMA with a spacing of 1µm between them. The growth program (Fig. 19e) was interrupted after 2 h. For further protection the NRs were infiltrated by spin-coated photoresist. In order to make their tips accessible for the scanning probe the polymer was partly etched back in oxygen plasma (Fig. 21a). The photoresist encapsulation is also beneficial for increasing the PFM contrast. Note, that in the arrangement of Fig. 21a the underlying ZnO seed will also contribute to the piezoresponse signal. However, due to its lower vertical dimension compared to the NRs, the majority of the applied voltage will drop across the NRs.

Nevertheless, for comparison I determined the d₃₃ constant for the seed layer as well before resist infiltration at a position far from the patterned region (Fig. 21b).

Figure 21. Schematic of the PFM characterization of ZnO NRs grown on PLD/Pt/sapphire (a) and the PLD/Pt/sapphire seed layer without the NRs (b). In order to avoid the fracture of the NRs in contact mode scanning thicker NRs were synthesized, which were infiltrated by spin-coated photoresist.

In order to study the impact of seed layer postdeposition annealing on NW alignment, I annealed three ZnO seed layers deposited by CVD on Si(100) in a quartz tube furnace at temperatures of 500 °C, 700 °C and 900 °C for 2h. The original thickness of the layers was

57

200 nm and the annealing atmosphere was artificial air (20% O₂ + 80% N₂). A fourth sample was left without heat treatment. After annealing ZnO NWs were grown on all four CVD ZnO layers by our template assisted wet chemical method, as described above. Although the shape of the air temperature profile in the oven was the same (Fig. 19e), in this case I applied 95 °C (instead of 85 °C) and a concentration of 10 mM (instead of 4 mM), and wider nucleation windows (~500 nm) were generated in the PMMA. The resulting NWs were examined in SEM. Before NW growth the surface of the seed layers was studied by AFM in semicontact mode. The deposition of CVD ZnO layers was done at CEA LETI (France).