Thin gold films are used for many applications such as in circuit boards and sensors manufacturing. Au is favourable because it is highly conductive and it is not easily oxidized. Nowadays there is special interest for the formation of self-assembled monolayers (SAMs) of thiolated molecules on gold surfaces; this is achieved by taking advantage of the covalent link between the sulfur of a thiol group of the molecule and the Au substrate. To facilitate the formation of SAMs, it is desired that the molecular axis is almost perpendicular to the Au surface. In the case of alkane thiols, the bonding configuration of the gold and the molecular axis is tilted 20°-30° off the surface normal only when the Au crystallites in the film are orientated in the (111) plane direction. Preparation of epitaxial Au(111) has primarily involved distinct methods all of which require ultra high vacuum conditions (around 10-10 mbar). However, other easier to implement techniques also exist to achieve the (111) plane direction of gold surfaces such as furnace annealing and flame annealing. In the former the heating rate is almost constant whereas in the later the heating rate is linear or exponential. In this project we study the crystallization and surface morphology of gold growth at room temperature on Si-based substrates after ex-situ furnace and annealing.
For instance, from the XRD analysis in Fig 1 we can deduct that after both furnace and flame annealing, Au thin films on SiO2 substrates can be crystallized despite most of the substrate being amorphous and containing only a small amount of SiO2 crystallites. Furthermore, the differences between the cubic cell parameters are around 4.08Å for gold (PDF No 89-3697 and 4.52Å for SiO2 crystals (PDF No 89-3609). Annealing overcomes these mismatches by arranging the Au layers in the (111) plane direction and changing the morphology of the surface. The fact that gold, and SiO2 have cubic structures seems to be the only condition for both elements to allow epitaxial growth. Both slow and rapid annealing gives ignorable background of the XRD spectra and hence clean and complete crystallization. Therefore crystallization of gold strongly depends on the temperature even if the sample is rapidly annealed.
After annealing at temperatures higher than the eutectic point of the Au/Si system, crystallized islands of different shapes and sizes were formed on the surface of the samples by both annealing techniques Figs 2 and 3. The mismatch between the cell parameters of the Au, Ti interlayer and SiO2 produces a strained Au layer during evaporation. Strain relaxation at higher temperatures would be responsible for the initial stages of the cluster formation. Above the eutectic point the layer melts. The melting starts most likely from the grain boundaries and from the interface. Moreover, during the quenching of the sample, Au atoms should crystallize from the liquid, when the temperature reaches the liquid curve in the phase diagram. This is done by growing the existing solid gold grains. As the temperature continues to fall, more and more Au crystallizes. Eventually, when the temperature drops to the eutectic point, the liquid crystallizes with the characteristic structure of lamellae islands. Nevertheless hexagonally shaped islands are formed, if the sample is annealed well above the eutectic temperature.
Fig 2. AFM analysis of Au(40nm)/SiO2 after furnace annealing.
Fig 3. AFM analysis of the sample Au(40nm)/SiO2 after flame annealing
The principal differences between both types of annealing techniques here are the heating (cooling) rate and the time of annealing. Furnace annealing is slower and provides more control for thermal conditions than flame annealing. Regarding the final surfaces, both types of annealing provide epitaxial gold islands with maximum mean crystallite diameters of around 90 nm. The annealing history seems to be more important for the final shape of the clusters. Because slow annealing rates are achieved by the furnace technique, the growth rate (G) can be related to the derivative of DT (Fig. 4(a)). From information given by XRD, AFM and SEM, we can deduce the morphology evolution of the sample surface during annealing. Three main events can be identified as seen in Fig. 4. The exact point at which each event happens is difficult to determine precisely because changes occur progressively. The dashed lines in the figure are rough estimates for each transition. The situation is harder to predict for the flame annealing than for the furnace annealing. Transitions during flame annealing occur abruptly due to the fast heating rates. Initially after evaporation the surface consists of a strained and amorphous Au layer which crystallizes after annealing. The slow increase of D (crystallites mean diameter) in the early annealing stages of growth by both techniques may be due to the highly disordered initial structure (excess point defects, dislocations, lattice strain, etc.) which is present in a thin film after deposition. Gold island nucleation sites may be formed in those disordered points facilitated by the relaxation of the strained Au layer, when the temperature and time of annealing increases. The clusters coexist with the gold layer substrate and they grow in size with the annealing temperature and time (mixed zones in the figure). The maximum inflection point of the derivative of DT is close to the eutectic temperature of the Au/Si system (363°C). This means that the nucleation rate is highest around this temperature. Annealing the sample at higher temperatures produces more gold that diffuses from the substrate to the clusters to form separated islands of different shapes (relaxed islands zones in the figure). Well above the eutectic temperature, the relaxed islands have hexagonally shaped borders.
Fig. 4.Crystallites mean diameters of Au/SiO2 following (a) furnace annealing and (b) flame annealing.
The fact that above 360°C the surface morphology transforms to gold islands spread over the SiO2 substrate is a good indication for obtaining an epitaxial Au(111) layer covering the whole of the SiO2 substrate. This could be achieved by simply annealing the sample at temperatures below 360°C. If the furnace annealing technique as presented here is used for this purpose suitable temperatures are in the range of 300°C to 310°C, just before DT changes drastically and islands stars to nucleate (Fig. 4(a)). However it is important to remember that the total crystallization of the strained Au layer is not obtained at that temperature range (Fig. 1(a)). The situation is different if flame annealing is used for the same purpose. Despite being much faster and easy to perform than the previous technique, it allows less control of the temperature. An epitaxial Au(111) layer, covering almost all the surface, could be achieved by annealing the sample for a few seconds (less than 30 s in which Dt changes drastically (Fig. 6(b)). In the furnace annealing technique the (111) crystallinity of the strained Au film increases but complete crystallization is not obtained (Fig. 1(b)). This project is being performed in collaboration with the University of San Marcos (Peru).
For further reading about this project:
Luis De Los Santos V., Dongwook Lee, Jiwon Seo, Lizbet Leon F., Angel Bustamante D., Seiichi Suzuki, Yutaka Majima, Thanos Mitrelias, Adrian Ionescu and Crispin H.W. Barnes, 'Crystallization and surface morphology of Au/SiO2 thin Films following furnace and flame annealing', Surface Science (2009) doi:10.1016/j.susc.2009.08.011