Sputtered gold films as catalyst
The first series of experiments was performed with Si and borosilicate glass substrates. Borosilicate glass was used in this case to prove that the silicon in the grown nanowires originated from the precursor and not from the substrate. A uniform, thin layer of gold was sputtered onto the substrates with a sputtering time of one minute, corresponding to a thickness of about 10 nm. After transfer of these substrates into a tube reactor, NPS was carried into the reactor by a stream of argon.
At a reaction temperature of 375 °C silicon NWs formed within a typical reaction time of 1 h. SEM images of the NWs show a thickness of 1 µm and varying length of up to several 100 µm (, left and center).
Figure 1 SEM images of NWs grown on gold-sputtered Si exposed to NPS vapor at 375 °C for 1 h (left: side view; center: top view) and 30 min (right: top view). In the insert of the center view, the smaller, branching structures on the NWs become visible. (more ...)
The NWs do not grow evenly but are buckled in multiple directions, which presumably indicates that the catalytically active gold nanoparticles are of irregular shape and thus excrete the silicon nanowires in different directions. Nevertheless, several single NWs growing in a straight fashion over several hundreds of micrometers on the surface of the NW layer could be observed. The surfaces of the formed NWs are covered with smaller structures indicating that the NWs may still contain gold atoms, which catalyze the “branching” (, insert in center image). Shortening the reaction time to 30 min resulted in a loosely packed layer of shorter NWs (, right). Backscattered-electron images and EDX mapping of the NWs show that their bases as well as their bodies basically consist of pure silicon, whereas their tips are enriched with gold (). These findings support the suggested VLS mechanism.
Figure 2 Analysis of NWs grown on gold-sputtered Si for 1 h at 375 °C. Top left: Backscattered-electron image. The bright spots at the tips indicate the presence of a high-z material (gold). Top right: Composite EDX mapping image, green = gold, red (more ...)
To learn about the structure of the NWs, additional analysis by TEM was carried out. The TEM measurements confirm the buckled structure as well as the branched surface of the NWs (, center). The HRTEM measurements show that the NWs are crystalline. The lattice constant of 3.2 Å, which can be seen in the fast Fourier transformed (FFT) image and the selected-area electron diffraction (SAED) pattern, as well as the hexagonal pattern visible in the latter indicate the presence of the cubic Si lattice.
TEM images of NWs grown on gold-sputtered Si for 1 h at 375 °C. The FFT of the HRTEM image and the SAED pattern indicate the crystallinity of the Si NWs.
The influence of the gas flow on the nanowire deposition was investigated by varying it between 0.1 and 1.0 L/min. In all cases NW growth was observed. At the lower flow limit, the length and diameter of the individual NWs resembled those of the NWs in the previously mentioned experiments; however, the substrate was not completely covered with NWs but rather showed only sparsely distributed, small NW islands. These results were similar to the ones obtained with a shortened reaction time and can in both cases be explained by the relatively small amount of precursor being carried into the system. At the upper limit of the gas flow, the NWs grew efficiently, covering most of the substrate. Nevertheless, a large amount of precursor seemed to pass through the system without the formation of Si, since at the exit valve, a white insoluble material formed in copious amounts. On account of these results, all other experiments were carried out with an argon flow that lay well between 1.0 and 0.1 L/min.
The reaction was repeated at 650 °C, but otherwise under the same conditions. SEM measurements of the formed NWs showed similarly shaped growth patterns as for the NWs formed at 375 °C (). However, with diameters of about 650 nm, the NWs were slightly thinner than in the previous experiments. One possible reason for this could be that at temperatures as high as 650 °C the sputtered gold layer disrupts to form smaller particles, enhancing the NW growth. Raising of the reaction temperature to as high as 900 °C resulted in the deposition of amorphous silicon throughout the reactor.
SEM images of NWs grown on gold-sputtered Si by exposure to NPS vapor for 1 h at 650 °C.
Sputtering and in situ formation of nanoparticles by dewetting
To produce more uniform layers of low diameter NWs, the sputtered Au layers were transformed into nanoparticles before the Si deposition. The amount of gold on the substrate surface was controlled by varying the sputtering time to obtain differently sized gold nanoparticles. Sputtering times were 30 s (sample 1), 1 min (sample 2) and 2 min (sample 3). By means of AFM measurements, we determined the deposition rate to be about 10 nm/min.
Upon heating these films to 375 °C no obvious changes occurred, while at 650 °C nanostructures formed within 1 h as a result of a dewetting process [28
]. As visible in the SEM images (), the size and shape of the nanostructures formed on the three kinds of samples varied considerably. While on samples 1 and 2 separate nanoparticles with diameters of 30–100 nm (sample 1) and 170–300 nm (sample 2) were found, sample 3 rather showed a network of gold after annealing (, right).
Figure 5 SEM image of gold nanoparticles formed on the native oxide surface of Si after annealing of Au films deposited by sputtering for different lengths of time (left: 30 s; center: 1 min; right: 2 min). The annealing was performed at 650 °C for (more ...)
When these substrates were treated with NPS vapor for 1 h at 650 °C (treatment “a”), the resulting NWs on the three samples also differed in their shapes and diameters. On sample 1a, a dense layer of long NWs with diameters varying between 100 and 200 nm could be observed (, left). Again, the NWs were buckled, as in the case of nonannealed Au layers. Additionally, bulkier structures and a few thicker NWs could be observed on the substrate surface. The NWs on sample 2a grew in a more straight fashion with a diameter of 600–700 nm (, center). In both cases the diameters of all NWs were more than twice the size of the nanoparticles. The NWs on sample 3a, on which no nanoparticles had been formed during the annealing step, were even bulkier than the ones on the other samples, with a diameter of around 800 nm or more (, right). Remarkably, these NWs showed significant branching.
SEM image of NWs grown at 650 °C for 1 h on Si surfaces with annealed gold films (left: sample 1a; center: sample 2a; right: sample 3a).
Since NW growth was achieved at temperatures as low as 375 °C on sputtered surfaces, the NW growth was repeated at this temperature for 1 h (treatment ”b”) by using again gold films previously annealed at 650 °C for 1 h, as described above. SEM measurements showed the growth of NWs with a diameter of 60–100 nm on sample 1b (, left). The NWs have a lot of kinks and are clustered together forming a dense layer on the substrate. On sample 2b, the NWs can be seen more clearly (, center). Their buckled growth resembles the NW formation on nonannealed substrates, while their diameter is somewhat smaller, between 300 and 500 nm. On sample 3b, the networked structure of the annealed gold layer is basically maintained. Structures with a size of several micrometers formed on the surface of the substrate. On top of those formations, agglomerated NWs with a diameter of 1 µm started to grow. In between these structures, the formation of particles with a size of around 500 nm could be observed. These particles contained silicon and a high amount of gold, as determined by EDX measurements. No NW growth could be observed from these particles (, right). At both temperatures, the diameter of the NWs exceeded the diameters of the catalytically active nanoparticles. Nevertheless, when the growth was performed at 375 °C, the diameter of the NWs was significantly smaller than the diameter of the NWs grown at 650 °C ().
SEM image of NWs grown at 375 °C for 1 h, on annealed gold films of different sputtering times (left: sample 1b; center: sample 2b; right: sample 3b).
Size distribution of the NWs obtained at different temperatures, grown on sputtered gold films of different thickness that had been annealed at 650 °C before the NW formation.
We assume that the nanoparticles become larger by taking up the silicon atoms from the precursor, which results in thicker nanowires. At higher temperatures, the alloy can take up more silicon, resulting in even bigger nanoparticles, further increasing the diameter of the extruded NWs.
Deposition of preformed nanoparticles from solution
To decrease the diameter of the NWs, gold nanoparticles with a size of 60 nm were synthesized by following standard protocols [42
] and deposited from solution onto Si substrates. For this, the native oxide layer of the silicon wafers was modified by a monolayer of 3-aminopropyl-terminated siloxane [43
], the amino groups of which are able to coordinate to the Au nanoparticles. The chemisorption of the nanoparticles proceeded by simple immersion into the respective solution and resulted in surfaces that were evenly, but not closely decorated by the nanoparticles (). The average distance between two nanoparticles could be estimated to be about 1 µm.
SEM image of the chemisorbed gold nanoparticles on the aminopropylated Si surface.
Using these surfaces without further treatment, no NW growth was observed below 650 °C. At this temperature, NWs with a diameter of less than 100 nm and a length of up to 10 µm formed on the silicon surface (, left). We assumed that at this temperature, the organic stabilizers thermally decomposed, making the surface of the nanoparticles accessible for the precursor. To cross-check this hypothesis, we used an alternative, but established protocol [44
], i.e., the removal of organic material by a treatment with H2
plasma prior to NW deposition. Indeed, after this treatment, the formation of NWs with a diameter of 60 nm and a length of up to 10 µm could be observed already at reaction temperatures of 375 °C (, right). Due to the relatively large distance between the nanoparticles, the NWs did not form a dense carpet on the substrate, as was the case with the sputtered substrates. The buckling of the NWs even when growing far apart demonstrates that the buckling does not arise from contact/steric hindrance within the more densely packed NW layers. While the images of the two different methods look quite similar at first sight, it should be mentioned that the pretreatment with H2
plasma and the low temperature seems to avoid the formation of the silicon nanoparticles that can be found in between the NWs on the sample formed at 650 °C.
SEM image of NWs grown from nanoparticles at 650 °C without prior treatment (left) and at 375 °C after H2-plasma treatment for 40 min (right). Observe that at the lower temperature no round Si particles could be found on the sample.
Nanoparticles from “liquid bright gold” as precursor
One alternative method for the deposition of thin gold films to surfaces is the use of “liquid bright gold”, also called “porcelain gold” or “gold ink”. This material, which is formed by heating gold dust with sulfur and terpenes, was commonly used in the manufacturing and refinement process of porcelain [45
]. When the tar-like material is painted onto ceramic surfaces, it can be converted to gold films by simple heating in air. Today it is typically applied by screen printing for the decoration of porcelain and glassware [46
To obtain thin gold films, commercial Screen Printing Brightgold 14603 (Surcotech) was diluted with dichloromethane and spin coated onto the native oxide layer of a Si substrate. At 7000 rpm a dense coating was achieved in the form of a sticky film. Annealing of this film at 650 °C for 1 h in the presence of air resulted in dense but inhomogeneous gold particle coverage. Those particles partially resembled the nanoparticle assemblies obtained from the annealing of the sputtered gold films, although their size and shape distribution was much wider (, right). Many disruptions as well as areas with more aggregated particles could be found on the substrate surface, which is presumably due to inhomogeneous drying/decomposition of the films during its thermolysis (, left and center).
SEM images with different magnifications of a spin coated film of “liquid bright gold” on Si after annealing at 650 °C for 1 h.
Again, NW growth was achieved by treatment with NPS at different deposition temperatures. At 375 °C NW growth could be observed, but the NWs were less densely packed than in the sputtered cases. In the background small particles could be found, which contained larger amounts of gold as shown by EDX measurements. The NWs were thin with many kinks, growing on the substrate in an insular mode (, left). At 650 °C and a deposition time of 1 h the observed NWs seemed to grow in a straighter fashion with a length of several 100 µm and a diameter of around 150 nm. Additionally, smaller-scaled NWs growing in a buckled way could be observed on the substrate (, right).
SEM image of Si NWs obtained from spin-coated “liquid bright gold” films after annealing at 650 °C, followed by treatment with NPS at 375 °C (left) and 650 °C (right) for 1 h each.
The polymeric nature of “liquid bright gold” offers opportunities for patterning and hence controlling the growth areas of the NWs. One well-established method for the patterned deposition of polymer films is microcontact printing [47
], which is generally considered advantageous since it is an “additive” method. In this particular case the need for dichloromethane or a similar solvent makes this process unsuitable due to the well-known swelling of the stamp material with these solvents [48
]. We figured that the gold in these resins might be reducible to the elemental state by irradiation and devised, therefore, a scheme similar to the well-established photolithographic process, which by definition is a “subtractive” one. When a spin-coated layer of the “liquid bright gold” was illuminated with a mercury-vapor lamp through a mask consisting of a metal pattern on a planar quartz substrate, no visible changes took place in the polymer layer. Nevertheless, after washing with dichloromethane, the nonirradiated areas were dissolved, while the irradiated parts of the “liquid bright gold” persisted, reproducing the pattern of the mask (, left). Obviously cross-linking occurs during the irradiation, decreasing the solubility of this material (negative photoresist). Annealing of this patterned film for one hour in the presence of air, as described above, left a thin but visible layer of gold with the same pattern (, center). The gold-patterned substrates were placed into the reactor and treated with NPS for 1 h at 650 °C (, right) and 375 °C. NW growth could be observed in both cases; however, the coverage of the substrate at 375 °C turned out to be not dense enough to form a satisfying pattern ().
Figure 12 Optical microscopy images of the different steps of the irradiation-induced pattern formation. Left: The “liquid bright gold” film after UV irradiation and washing with dichloromethane. Center: The same area after annealing at 650 °C. (more ...)
SEM images (different magnifications) of Si NWs obtained from UV-patterned spin-coated “liquid bright gold” films after annealing at 650 °C, followed by treatment with NPS at 375 °C for 1 h.
At a reaction temperature of 650 °C the pattern was much more pronounced. SEM images of the substrate showed that long NWs grew within the previously irradiated areas, similar to those in , right. These NWs have a diameter of around 800 nm and grow in a straight fashion over several micrometers (). Although the borders of the pattern were relatively sharp, a zone about 100 µm wide could be observed in which the NW growth differed from inner parts of the NW-covered areas. The NWs at the border zone had a diameter of up to 2 µm and grew longer than on the rest of the pattern (, left, insert). It remains unclear, why these zones of different morphology are formed in the first place. Currently, investigations are under way to understand this peculiar behavior.
SEM image of the border region of a Si NW pattern obtained from UV-patterned “liquid bright gold” films after annealing at 650 °C, followed by treatment with NPS at 650 °C. Left: Top view, right: Side view.