4.3 Results and discussion
5.3.1 Layer thickness calibration
In order to lower the thermal conductivity of the samples, it is desirable to increase the inter- face density as much as possible by targeting the thinnest attainable layers of the constituent materials. It has also been suggested that, in this type of superlattice, changing the super- lattice period can be used to tune the centre frequency of the phononic bandgap [10, 18]. As such, individual spin-coated layers were calibrated to find the obtainable layer thicknesses. The layers were measured through surface profilometry, and laser ellipsometry as described previously.
The thickness of the PMMA layers was varied by changing the spin-speed of the spincoater between 500 and 9000 rpm, at the same time, the duration and acceleration were held constant at 30 seconds and 5040 rpm/s. The results of this layer optimisation process are shown in Figure 5.9 a). Initially, the Ludox layers were measured with surface profilometry, however the nature of the nano-particle material on the substrates was such that reliable steps in the material could not be easily created. Instead, as also done by Schneider et al. [10], the nano-particle layer thickness was measured with ellipsometry. This method was used to develop a relationship between the layer thickness and the nanoparticle concentration in the spin-coating solution. The ellipsometry process was carried out twice on each sample
a)
b)
0 2,000 4,000 6,000 8,000 50 100 150 Spin-speed (rpm) La ye r thic kne ss ( nm ) 2.2% (w v) 0 50 100 150 100 200 300Spin solution concentration (Lg)
La ye r thic kne ss ( nm )
Figure 5.9: Plots showing the layer thickness calibration of both the PMMA and Ludox (SiO2 nanoparticle) layers. The PMMA layer thickness was predominantly controlled by set-
ting the spincoater speed. In contrast, the nanoparticle layers were adjusted by controlling the spin-coater solution. a)The PMMA spin-profile was calibrated over the range 500-9,000rpm. This resulted in layers varying from 175 nmdown to 65 nm. b) The ellipsometry results for the Ludox nanoparticle material. This result demonstrates that the nanoparticle materials could be controlled with solution concentrations down to approximately 80 nm thickness.
at 50 and 70 angles of incidence. The Ludox layer thickness was adjusted by varying the solution concentration, while the spin speed was held constant at 5000 rpm. The Ludox concentrations tested were 30, 40, 80, 130, 180 g/L. The samples were baked at 100 C for 15 minutes on a hotplate to evaporate the solvents. The results of the Ludox ellipsometry thickness measurements are shown in Figure 5.9 b).
The thinnest PMMA layer attainable was 65nm(6000 rpm, 30s) as an individual layer. This thickness may vary when coated in the superlattice morphology. This is due to the surface roughness and porosity of the Ludox layers in which the PMMA will be deposited on top of. These parameters may alter the disjoining pressure and evaporation rate of deposited medium [161], and create a di↵erent equilibrium thickness spin-speed relation [151]. The thickness of the Ludox layers could be controlled down to 80 nm using spin-solution con- centration (nanoparticle concentration of 30 g/L). However, there was substantial variation in the thickness measured depending on the sample area tested. It was found that at high spin-speeds, comet streaking in the films was significant, and the resulting superlattice were poor overall quality. The cometing e↵ects were much less severe under 6000 rpm however.
As already eluded to, the applicability of using ellipsometry as a thickness measurement technique for nanoparticle materials is uncertain. Null-point laser ellipsometry only models two interfaces (ambient–film, film–substrate [154]), whereas a nanoparticle film will actually be composed of multiple other interfaces between adjacent nanoparticles and also between nanoparticles and the air. There are alternative methods which are able to measure such a film, and even estimate porosity from the measurements [157], however they were not available for this work.
Due to the variability of the spin coating results, it is noted that the layer thickness of these layer can only be coarsely targeted [10]. The final thicknesses of the superlattices and individual layers is measured in-situ by SEM cross-section.
From this work, it was concluded that the optimal recipe for these structures was: PMMA was spun at 5000 rpm for 30s. The layer is then baked at 100 Con a hotplate for 15 minutes. The Ludox spin solution was diluted to a nanoparticle concentration of 30 g/L and spun at 5000 rpm for 30 s. This layer is then also baked at 100 C on a hotplate for 15 minutes. This process is repeated until 10 total bilayers have been deposited.
Post measurement, several SEM images were taken of the PMMA-SiO2 superlattices in cross- section. These can be seen in Figure 5.10, where the dark layers indicate the PMMA, and the lighter layers show the silicate nanoparticles. In general, SEM micrographs allowed for more precise measurement of the superlattice thicknesses. To obtain the average superlattice period of these samples, three separate superlattices samples were analysed. The thickness of every layer was measured and averaged. The average SiO2 nanoparticle layer thickness from these samples was 65nmwith = 7.68nm(n=30), while the average PMMA layer thickness was 18.5 nm with = 4.7nm (n=30). Interestingly, the PMMA layers in multilayer were much thinner than those measured as individual thin films previously.