The controlled localized bottom-up synthesis of silicon nanowires on arbitrarily shaped surfaces is still a persisting challenge for functional device assembly. In order to address this issue, electron beam and focused ion beam-assisted catalyst deposition have been investigated with respect to platinum expected to form a PtSi alloy catalyst for a subsequent bottom-up nanowire synthesis. The effective implementation of pure platinum nanoparticles or thin films for silicon nanowire growth has been demonstrated recently. Beam-deposited platinum contains significant quantities of amorphous carbon due to the organic precursor and gallium ions for a focused ion beam-based deposition process. Nevertheless, silicon nanowires could be grown on various substrates regardless of the platinum purity. Additionally, p-type doping could be realized with diborane whereas n-type doping suppressed a nanowire growth. The rational utilization of this beam-assisted approach enables us to control the localized synthesis of single silicon nanowires at planar surfaces but succeeded also in single nanowire growth at the three-dimensional apex of an atomic force microscopy tip. Therefore, this catalyst deposition method appears to be a unique extension of current technologies to assemble complex nanowire-based devices.
Silicon nanowires (SiNWs) are one-dimensional semiconductors commonly proposed as future building blocks for advanced field-effect transistors,1 sensor devices,2 and even for efficient energy conversion applications.3 Common techniques for bottom-up SiNWs growth are the so-called vapor-liquid-solid (VLS)4 and the vapor-solid-solid (VSS)5 synthesis using metals, such as Au, Al, and Pt as catalysts.6,7 Although various functional nanowire devices have been shown so far, efficient device assembly strategies with accurate nanowire localization and alignment are still a major challenge. A straightforward approach is the pre-patterning of the catalyst material using photo and electron beam lithography. Also nanosphere lithography, masking with porous materials, lateral manipulation of catalyst nanoparticles,8 and local substrate heating to provoke local nanowire growth9,10 have been successfully demonstrated. However, each of these methods exhibits intrinsic limitations. Photo and electron beam lithography allow large-scale user-defined patterning, but they are constrained to rather planar surfaces due to the need of a resist layer. Nanosphere lithography or masking by porous templates can be readily applied, but they are not suitable for forming arbitrarily shaped patterns. The manipulation of catalyst nanoparticles and local substrate heating allow nanowire growth on nearly any object, but the technological realization is complex and time-consuming.
Based on these considerations, the localized platinum catalyst deposition was investigated by both electron beam-induced deposition (EBID) and ion beam-induced deposition (IBID) using a dual beam system, Quanta 3D FEG (FEI Netherlands). The maskless beam-induced deposition approach enables, in principle, arbitrary catalyst patterning similar to lithography but allows the modification of three-dimensional objects as well as planar substrates. Furthermore, IBID Pt has already been used successfully as catalyst for GaN nanowire growth.11
For the beam-induced deposition of platinum, gaseous (methylcyclopentadienyl)trimethyl platinum is used as organometallic precursor. It is introduced into the specimen chamber through a gas nozzle positioned in close proximity to the specimen surface, where the precursor adsorbs to the substrate (Fig. 1(a)). Either a focused gallium ion beam or an electron beam can be employed to decompose locally the adsorbed gas molecules leading to material precipitation and the release of volatile organic compounds. Due to the high content of carbon in the precursor, the deposited material is a Pt-C composite rather than pure Pt with a carbon content of approximately 55–65 at. % (Ref. 12) and 70–80 at. % (Ref. 13) for IBID and EBID, respectively. Furthermore, in case of IBID, also implanted Ga ions (5–18 at. %) are co-deposited.12 With respect to the SiNW synthesis, carbon as well as gallium may have an effect on nanowire growth. In order to study these effects, comparative control experiments were performed using a high purity Pt catalyst deposited by thermal electron beam evaporation.
(a): Schematic illustration of the beam-induced Pt deposition process with (1) scanning electron or ion beam, (2) inlet gas nozzle for the Pt-precursor, and (3) the substrate with Pt-C patterns or thin films, (b) and (c) show the PtSi droplet and island formation from a Pt-C IBID thin film (50 μm × 50 μm, nominal deposition height 2 nm, ion current 30 pA at 30 kV) deposit on (b): (100)-Si and (c): (111)-Si. The insets show the corresponding SiNWs after the growth process. On (111)-Si, island-like structures as well as small droplets (diameter 4 ± 1 nm) are present, whereas on (100)-Si only droplets with diameter 8 ± 3 nm are visible. The scale bar for the image and the inset is 250 nm and 5 μm, respectively.
(a): Schematic illustration of the beam-induced Pt deposition process with (1) scanning electron or ion beam, (2) inlet gas nozzle for the Pt-precursor, and (3) the substrate with Pt-C patterns or thin films, (b) and (c) show the PtSi droplet and island formation from a Pt-C IBID thin film (50 μm × 50 μm, nominal deposition height 2 nm, ion current 30 pA at 30 kV) deposit on (b): (100)-Si and (c): (111)-Si. The insets show the corresponding SiNWs after the growth process. On (111)-Si, island-like structures as well as small droplets (diameter 4 ± 1 nm) are present, whereas on (100)-Si only droplets with diameter 8 ± 3 nm are visible. The scale bar for the image and the inset is 250 nm and 5 μm, respectively.
The bottom-up SiNW synthesis was carried out from localized Pt-C spots and dewetted Pt-C films to realize single or areal SiNW growth using single crystalline silicon, silica, and sapphire as substrates. Sapphire was chosen to examine the role of silicon-based substrates for a Pt-silicide formation and the nanowire growth due to the fact that thermally induced and Pt-catalyzed SiNW growth was observed on a silicon substrate acting solely as Si source.14
For the SiNW synthesis, a VLS-like gas phase process based on the work of Baron et al. was used.15 The process was carried out in a quartz-tube furnace and commences with a pre-heating step at 800 °C for 80 min with a chamber base pressure of approximately 0.1 mbar, immediately followed by a cooling down to the process temperature of 700 °C. SiNW growth was realized with a constant growth time of 1 h at a total pressure of 100 mbars using a hydrogen flux of 270 sccm, and a 2%-SiH4/He flux of 30 sccm. These process parameters were utilized for all subsequently presented experiments, regardless of the catalyst formation or substrate type. In general, the catalyst dimension and morphology is a key parameter for the NW synthesis using a VLS-like process.15,16 In order to study the catalyst in the beginning of the process, substrates with pure Pt and Pt-C catalyst films were removed from the tube furnace directly after the pre-heating step of 800 °C. Inspections showed that both pure Pt and Pt-C exhibit a similar behavior for all studied substrate materials with respect to an observable droplet formation as known for Pt and ascribed to a PtSi formation.15–17 However, the number and size of droplets is in case of beam-deposited Pt-C about 60% smaller than for evaporated Pt. This observation fits the aforementioned Pt concentration of only 30%–40% and the droplet size dependency on the deposited total Pt volume.15 Furthermore, the result indicates that a significant amount of carbon must be removed during the pre-heating phase. A straightforward explanation would be the formation of volatile carbonaceous species at elevated temperatures similar to the results observed in Ref. 13, but here under vacuum conditions. The removal of carbonaceous species is also supported by the low miscibility of C in Pt with less than 1 at. %.18 Besides the Pt volume, also the substrate material and its crystal orientation play a significant role in droplet formation as shown in Figs. 1(b) and 1(c). While on (100)-Si (Fig. 1(b)), silica, and sapphire Pt droplets with a diameter of approximately 8 ± 3 nm are predominantly formed, (111)-Si leads to significantly smaller droplets of only 4 ± 1 nm in diameter but promotes additionally the formation of planar spacious island-like structures as shown in Fig. 1(c). Baron et al. reported on (100)-silicon samples, which were pre-annealed in a hydrogen atmosphere forming PtSi droplets with a diameter of about 45 nm.15
If the samples are exposed to the entire process, SiNW growth can be observed similarly within the range of experimental uncertainties for any substrate material (cf., Figs. 1(b) and 1(c) insets). Additionally, the SiNW growth on silicon-free sapphire rules out the possibility that the substrate promotes the SiNW growth rather than the SiH4 and that the droplet formation requires a PtSi phase formation. This is supported by the fact that for Pt nanocrystals melting temperatures as low as 600 °C were reported.19 As discussed in Ref. 20, in case of IBID implanted Ga ions may also induce the growth of SiNWs at temperatures as low as 400 °C–500 °C. However, SiNW growth was not observed at 700 °C for any region just irradiated with the focused Ga ion beam. As shown in the insets of Fig. 1, two appearances of nanowires can be observed: (i) highly tapered nanowires with small tip diameters below 50 nm and (ii) less tapered nanowires with large diameters in the range of 500 nm. While the highly tapered nanowires appear to be in good agreement with a VLS-like growth from Pt droplets as described in Refs. 15 and 16, the less tapered nanowires are neither aligned with the average droplet size of less than 10 nm nor with their distribution. Interestingly, a correlation between the approximate sizes of the island-like structures for (111)-Si was observed but the role of these islands is not clear at this point. As shown by Wawro et al.21 planar epitaxial PtSi tiles of about 70 nm in diameter create these islands on (111)-Si but their formation is strongly dependent on the Pt amount. For the beam-assisted deposition, the thickness depends among other things on the local precursor concentration. As the distance between sample and gas nozzle significantly effects the precursor concentration, a gradient in Pt-C film thickness is intrinsically created (cf., Fig. 1(a)). In order to study the SiNW growth in dependence on the initial film thickness or the total Pt volume respectively, a Pt-C chessboard pattern with varying thickness due to the varying distance to the gas nozzle (Fig. 2(a)) was deposited by IBID (ion current 30 pA at 30 kV) on a (111)-Si substrate. Prior to the deposition, the surface oxide was removed with buffered hydrofluoric acid. Each chessboard pattern consisted of 8 × 8 alternating squares with 5 μm in length. The “white” squares are Pt-C, whereas the “black” squares represent the Si substrate. The square in the upper right corner (Fig. 2(a)) showed the lowest film thickness due to the largest distance to the gas nozzle with an average thickness of 3.7 nm determined by atomic force microscopy (AFM). Consequently, the largest thickness was achieved for the lower left corner with a value of 6.0 nm.
(a): Chessboard pattern consisting of Pt-C (“white” squares) on silicon covered by native oxide (“black” squares). The average Pt-C thickness is indicated for each square in nm. The scale bar for (a) and (b) is 10 μm. (b): The same area after SiNW growth. While the SiNW length is significantly affected by the Pt-C thickness the overall diameter appears constant. (c) FIB cross section after synthesis, which shows evidently that the growth of large diameter SiNWs is far more complex than a common VLS synthesis indicated by the porous overgrowth beneath the nanowires. The scale bar is 1 μm.
(a): Chessboard pattern consisting of Pt-C (“white” squares) on silicon covered by native oxide (“black” squares). The average Pt-C thickness is indicated for each square in nm. The scale bar for (a) and (b) is 10 μm. (b): The same area after SiNW growth. While the SiNW length is significantly affected by the Pt-C thickness the overall diameter appears constant. (c) FIB cross section after synthesis, which shows evidently that the growth of large diameter SiNWs is far more complex than a common VLS synthesis indicated by the porous overgrowth beneath the nanowires. The scale bar is 1 μm.
Figure 2(b) confirms that SiNW growth occurs only at the Pt-C squares approving that Pt acts as catalyst. However, a preponderance of large diameter nanowires is present. A pronounced Pt-C thickness dependency with respect to the NW length but interestingly not to the SiNW diameter (approximately 450 ± 50 nm) is observable.
Baron et al. reported only on highly tapered nanowires with a tip diameter in the range of the initial droplet size (about 45 nm), but the catalyst was a flat PtSi cap rather than a spherical PtSi droplet.15 Focused ion beam (FIB) cross sections, depicted in Fig. 2(c), reveal that the growth mechanism for these NWs with large diameter appears far more complex than the growth observed in a common VLS-like nanowire synthesis. As shown in Figure 2(c), a highly porous substrate overgrowth was observed beneath the nanowires. A thorough discussion of this growth mechanism is beyond the scope of this letter and requires further investigations. Nevertheless, silicide formation appears as a crucial factor determining the nanowire growth. The PtSi phase exhibits the highest Si concentration and represents therefore the most likely catalyst composition able to promote SiNW growth if oversaturated in agreement with Ref. 15.22 While Refs. 15 and 16 show evidently a PtSi catalyst as flat cap or distinct droplet, respectively, TEM studies in combination with elemental analysis (data not shown) could not confirm Pt at the tip of polycrystalline large-diameter SiNWs.
The observation that the initial Pt droplet size is not related to the large nanowire diameter, while the film thickness significantly governs the growth rate, indicates a dynamic formation of larger PtSi catalyst structures at the surface. Such structures may be similar to the observed PtSi island-like structures. As shown here, the SiNW synthesis is independent on the nature of the substrates, hence their formation must be triggered by SiH4 supported by the surface thermo-migration ability of Pt-Si droplets.23 Assuming a critical catalyst size, thicker Pt layers would promote their formation and therefore an earlier onset of the SiNW growth.
In comparison with the aforementioned observations of areal SiNW growth, Baron et al. achieved a growth of single tapered nanowires from the individual PtSi droplets of about 45 nm in diameter.15 Based on the so far achieved results, separated Pt-C spots with 80 nm in diameter and 50 nm in height were deposited using EBID. For high-resolution patterning, depositions were made in a focused ion beam pre-patterned cavity on (111)-Si substrate (current 6.7 pA at 5 kV). In agreement with Ref. 15, also single SiNWs were obtained but with a larger diameter of 200 nm (inset of Fig. 3). Understanding the deviation of the nanowire diameter from the droplet size would require a detailed analysis of the catalyst morphology.
The growth of single SiNWs was realized from single Pt-C spots placed in a cavity (inset). By depositing Pt-C spots locally at the tip of AFM probes, SiNWs could be locally grown at the apex. The scale bars refer to 1 μm.
The growth of single SiNWs was realized from single Pt-C spots placed in a cavity (inset). By depositing Pt-C spots locally at the tip of AFM probes, SiNWs could be locally grown at the apex. The scale bars refer to 1 μm.
To finally demonstrate the proposed versatility of EBID or IBID catalyst formation, the apex of a commercial silicon AFM probe was used as substrate for SiNW growth. The probe tip was milled into a frustum by FIB leaving a plateau of 650 nm edge length. On this plateau, a Pt-C spot with 100 nm in diameter and a height of 50 nm was deposited by IBID as discussed before. As shown in Fig. 3, SiNW growth can be locally obtained at the apex of the truncated tip. SiNWs have grown in the vertical direction, as expected, enhancing potentially the aspect ratio of the AFM tip. To grow only one single SiNW requires further optimization. Other SiNWs are located at the lateral facets of the tip, probably grown due to residues of deposited Pt on the side walls oriented towards the gas nozzle.
As the majority of SiNW devices are based on their electrical properties, silicon doping must be additionally addressed to extend the scope of the approach. Doping of SiNWs was examined by reducing the silane flow to 20 sccm and adding 10 sccm of either phosphine (PH3, 200 ppm in He) or diborane (B2H6, 100 ppm in He) to achieve n- or p-type doping, respectively. While nanowire growth is obtained with pure SiH4 or with SiH4/B2H6, PH3 admixture to SiH4 seems to suppress completely the SiNW synthesis for both thermally and beam-deposited Pt. The origin of this observation is still under investigation. Intrinsic and p-doped SiNWs with diameters ranging from 150 to 250 nm were electrically integrated with microfabricated contacts. The measurements revealed an average resistance per length of 500 ± 400 MΩ/μm for the undoped and 8 ± 5 MΩ/μm for the p-doped SiNWs. Despite the significant deviations, the decrease in resistance of almost two orders of magnitude indicates the active incorporation of boron in the silicon lattice during the SiNW growth.
In summary, we have shown that beam-induced deposition of Pt offers the potential to act as a versatile catalyst for SiNW synthesis at planar but also at deliberately selected locations and fragile substrates. In presence of diborane during the growth process, p-type-doping can be obtained as well. Despite the relative high carbon content and implanted gallium ions of the IBID process, comparable results to thermally evaporated platinum were achieved. However, the overall growth mechanism, with respect to the observed large diameter SiNWs, requires further investigations. In order to demonstrate the unique capability of the shown approach, SiNWs were grown at the apex of an AFM probe.
We kindly acknowledge the Focused Ion Beam Center UUlm, which is supported by FEI Company (Eindhoven, Netherlands), the German Science Foundation (INST40/385-F1UG), and the Struktur- und Innovationsfonds Baden-Württemberg. In addition, we thank Dr. J. Biskupek from the Electron Microscopy Group of Materials Science at Ulm University for TEM investigations. The financial support by the German Federal Ministry of Education and Research (BMBF-NanoMatFutur: 13N12545) is gratefully acknowledged.