AVS2012 Session EM+NS-FrM: Low-Resistance Contacts to Nanoelectronics

It has been suggested that the phenomenon of Coulomb blockade could be achieved by placing a metallic nanoparticle between two tunnel junctions. While the Coulomb blockade has been well established theoretically and demonstrated on metal substrates¹, it is more challenging to observe on semiconductor surfaces due in part to the defective nature of the interfaces and to the depletion layer. We present an experimental study of two ultra small-capacitance normal tunnel junctions connected in series between the Si substrate and a STM tip. To achieve such a structure, we use an amine-terminated self-assembled monolayer (SAM) grafted on silicon (111) as the insulator layer, acting as a linker to attach gold nanoparticles on the surface. The SAM layer is grafted directly on oxide-free silicon through a Si-C bond formation using hydrosilylation reactions and is characterized by a low interface state density². Moreover, this SAM layer provides a long-term passivation (weeks) of the interface that prevents oxidation of the substrate during Au nanoparticle deposition. The SAM quality is characterized using an extensive range of techniques, including in-situ IR spectroscopy, spectroscopic ellipsometry, X-ray photoelectron spectroscopy (XPS) and Atomic force microscopy (AFM). The second capacitance is formed by the gap between the gold nanoparticle and the tip of the Scanning tunneling microscope/spectroscope (STM/S). The current-voltage measurements have been performed in ultra high vacuum. Several parameters have been investigated: silicon doping level, sample temperature, and size of the gold nanoparticles (AuNPs), ranging from 1 to 15 nm. The junction is achieved by either grafting synthesized AuNPs or depositing evaporated gold directly on the SAM. Preliminary data confirm that coulomb staircases are observed under different conditions, mostly clearly for highly doped substrates at low temperature (10K). The steps width and height of these Coulomb staircases depend on particle size. These results are an important step toward future control for single electron transistor and flash memory applications.

[1] Zhang, H.; Yasutake, Y.; Shichibu, Y.; Teranishi, T.; Majima, Y., Tunneling resistance of double-barrier tunneling structures with an alkanethiol-protected Au nanoparticle. Phys. Rev. B, 72, (20) (2005)

[2] D. Aureau, Y. Varin, K. Roodenko, O. Seitz, O. Pluchery and Y. J. Chabal, Controlled Deposition of Gold Nanoparticles on Well-Defined Organic Monolayer Grafted on Silicon Surfaces. . Phys. Chem. C, 114 (33), pp 14180–14186 (2010)

Fullerenes and other small organic molecules are used in organic solar cells, organic LEDs and molecular electronics system, and the interface between the organic layer and the metal electrode is critical to achieve the desired functionality. The majority of studies focusses on the interaction of molecules with metal surface, and the interaction of metals with organic surfaces has garnered much less attention. However, the addition of metal to an organic layer surface has been one of the bottlenecks in the fabrication of molecular electronics devices. We therefore present here a comprehensive study of the metal interaction with fullerene surfaces. Our past research has investigated the deposition of Au and Si on fullerene surfaces, and our presentation here focuses on the interface to transition metals Vanadium and Tungsten. All of these systems show a dynamic behavior: the metal atoms are highly mobile and thus perturb the C₆₀ matrix substantially.

The fullerene and metal atoms/films are deposited by electron beam and thermal evaporation, and the interface formation is observed with STM under UHV conditions in a sequential manner. V immediately diffuses into the fullerene matrix, and surface clusters are sparse. The STM images reflect the change in the local electronic structure of the molecules through the interaction with sub-surface V: the apparent height of molecules in contact with V is reduced, and their rotation ceases and the molecular orbitals can be identified by the characteristic C₆₀ substructure within the molecule.

We suggest that the subsurface V forms complexes with C₆₀ where charge donation to the fullerene cage occurs, and preferential bonding to the hexagonal face determines the molecule orientation. The increase of V concentration leads to agglomeration of V-clusters and consequently the extension of regions with a smaller apparent height in the filled state images. The charge exchange between metal clusters and fullerene matrix allows to observe the V-cluster growth within the matrix. The empty state images are essentially flat, and show small variations in topography and cracks in the fullerene layer for large V-concentrations (~ 0.6 to 1 ML). We will present a comprehensive model for the diffusion of V through the matrix, the complex formation and cluster growth. The behavior of W is distinguished by a larger percentage of surface clusters, and the dynamics of cluster formation within the matrix will be compared to V. However, both transition metals do not destroy the C₆₀ matrix, but only react to form carbides at elevated temperatures.

Optical properties of metals are less well known than those of insulators and semiconductors, because it is hard to achieve similar purity and crystallinity in metals. Many metals are reactive and easily form oxides, or they exhibit significant surface roughness. We report the dielectric function and optical conductivity of Ni-Pt alloys as a function of composition (10 to 25 atomic % Pt) from 0.8 to 6.5 eV. Our films are 10 nm thick and were prepared by physical vapor deposition (co-sputtering from pure Ni and Pt targets). To avoid reaction between Si and the metal alloy, films were deposited on thick thermal oxides (220 nm). Some films were annealed at 500°C for 30 s. Similar Ni-Pt alloys are used as Ohmic contacts in CMOS device processing, to achieve highly stable low-resistance contacts between copper back-end metallization and front-end silicon transistors. Our results will enable in-line process control of Ni-Pt alloy deposition using spectroscopic ellipsometry.

Since our metal thickness is below the penetration depth, the interference from the thick SiO₂ layers creates artifacts when extracting the optical conductivity. We minimize this issue by acquiring the ellipsometric angles over a broad range of incidence angles (20° to 80°), which varies the optical path length and thus shifts the interference problems to different energies. Our resulting dielectric functions are similar to those tabulated by Palik for pure Ni. The data are dominated by a Drude divergence due to free carriers at low photon energies. We can remove this divergence by multiplying with the photon energy. We find several trends: (1) The optical conductivity of the annealed films is always greater than that of the as-deposited films, due to improvements in crystallinity and reduced grain boundary scattering after annealing. (2) All four alloys show conductivity peaks near 1.5 and 4 eV due to transitions from the d-like valence bands to the s-like conduction bands. (3) These peaks are significantly broader and weaker than those in pure Ni, but at the same energy. The broadenings increase with increasing Pt content. However, the amplitude of the 4 eV conductivity peak remains constant near 3500/Ωcm, independent of Pt content.

From electronic structure calculations for pure Ni and Pt and a Ni₃Pt ordered compound, we find that Ni, Pt, and Ni-Pt d-bands have similar energy, which explains why the 4 eV peak in the conductivity does not shift with Pt addition. Furthermore, the bandwidth of the Ni 3d bands is smaller than that of the Pt 5d bands, consistent with the increase in the broadening of the optical transitions.

This work was supported by NSF (DMR-11104934).

Secondary Ion Mass Spectrometry (SIMS) has shown great resilience over the last decades in keeping up with the aggressive downscaling of advanced CMOS technology. Major improvements contributing to the staying power of SIMS are lower primary ion beam energies to meet the ever more stringent depth resolution demands [1] and application of novel external standard-free calibration methods for quantification in the near-surface region [2]. Here we demonstrate state-of the-art applications of SIMS to Ultra-Shallow Junction (USJ) formation and in-situ doped thin epitaxial layers.

We present As-USJ extension formation for nFET with junction depths below 120 Å, obtained using low energy ion implantation and micro-second flash (μ-flash) annealing. SIMS depth profiling employing a 200 eV Cs⁺ beam provides detailed information on diffusion and segregation of As at the sub-nm scale for different annealing conditions. Low energy implantation of P has been proposed as an alternative to As for the formation of Source/Drain (S/D) regions to reduce crystalline damage. This is crucially important for advanced CMOS technology based on Extremely Thin SOI or FinFET. Here, we present different analytical approaches to determine the most accurate quantification for shallow P concentration profiles in Si. Also, we have employed 3D atom probe tomography to independently determine in-depth [P] profiles for SIMS calibration purposes [3].

For pFET processes, nm-scale control of B-diffusion is instrumental to obtain highly activated and abrupt B-USJ. Here we present a novel doping strategy employing ultra-thin solid source Si:B diffusion sources –in combination with μ-flash annealing– to form the B-USJ extensions. Presence of high [B] and minimal diffusion length necessitates use of ultra low O₂⁺ impact energy for accurate determination of junction depth and abruptness. Regarding S/D formation, In-Situ Boron Doped (ISBD) SiGe is an important technology element for pFET strain enhancement. Quantitative analysis of [B] in SiGe using reactive low energy O₂⁺ ion sputtering is complicated due to large yield variations as a function of [Ge] [4]. We present a calibration protocol based on multiple B-implanted epitaxial Si_1-xGe_x standards on Si(100) with constant [Ge] ranging from 20 to 50 at.%. This approach allows for explicit correction of both SiGe sputter yield and B⁺ and Ge⁺ yield variations as function of [Ge], enabling quantitative analysis of ISBD SiGe.

[1] A. Merkulov et al., JVST B 28(1) (2010) C1C48.

[2] W. Vandervorst et al., AIP Conf. Proc. 931(1) (2007) 233-245.

[3] M.J.P. Hopstaken et al, SIA, DOI 10.1002/sia.4916 (2012).

[4] Z. Zhu et al., SIA 43(1-2) (2011) 657-660.