SIMS2015 Session SR1-ThM: Ultra High Spatial Resolution SIMS

One of the important growth drivers for the semiconductor industry is the shrinking of transistors to smaller dimensions. System performance and functionality increase as the density of transistors which can be integrated onto a single chip increases. To maintain the device performances for very small dimensions, additional improvements are needed such as the implementation of new materials (high-k’s, metal gate, III-V or group IV alloys) and the introduction of new 3D device architectures, e.g. FinFET’s.

In this context, the metrology needs (determination of the exact composition and thickness of deposited/grown thin films) are more and more shifting from the analysis of blanket samples to more real devices where thin films are grown in very confined volumes (with dimensions below 20 nm). Such metrology step makes the standard analysis methods like XPS, SIMS and RBS no longer applicable due to a lack of spatial resolution. The concept of Self Focusing SIMS (SF-SIMS) overcomes the spatial resolution limitations of SIMS without sacrificing the sensitivity [1]. The concept is based on determining the composition of a specific compound using cluster ions which contain the constituents of the compound. Their formation mechanism implies that all cluster constituents originate from the same collision cascade and are emitted in close proximity (< 0.5 nm). As such, the composition information becomes confined (i.e. self focused) to the areas where all constituents are simultaneously present.

Recently, we demonstrated the application of SF-SIMS to quantify the Ge content of SiGe films grown in narrow trenches [1,2]. In the present work, the SF-SIMS approach is applied on III-V materials. In a first part, we will show that the In content of InGaAs films grown on InP/Si in STI (Shallow Trench Isolation) could be accurately determined for very confined volumes. The obtained results will be compared to analyses performed with other complementary methods such as AES, TEM, Atom Probe Tomography, .... Attempt to explain the shape of the recorded depth profiles will be made by comparing them to the real shape of the trenches obtained by x-section TEM. In a second part, the SF-SIMS approach will be applied to determine the In concentration in each layer of more complex stacks, i.e. InGaAs/InAlAs/InP/Si, in which all dimensions, i.e layer thickness and trench width, are of the order of 20 nm.

[1] A. Franquet et al., “Self Focusing SIMS : Probing Thin Film Composition In Very Confined Volumes” in preparation.

[2] A. Franquet et al., “Quantification of Group IV Alloys in Confined Structures: the Self Focusing SIMS Approach”, SIMS Europe 2014.

Inductively coupled plasma ion sources are now being used to generate both positive and negatively charged primary ion beams for SIMS. The high brightness, low energy spread, long life and beam current stability are all factors that have provided enhanced oxygen focused ion beams, across a range of beam energies.

Here we present characteristic performance data for the Hyperion II ion source, when operated in both ion polarities. A combination of the increased brightness and reduced energy spread, when compared to a duoplasmatron, has resulted in a factor of 4 gain in SIMS imaging resolution. The Lawrence Livermore National Laboratory (LLNL) NanoSIMS50, has been the first instrument to demonstrate sub-50nm SIMS imaging resolution, when operated with a Hyperion II ion source. Here we present image and spot size data from the LLNL instrument, as well as from our higher energy oxygen focused ion beam (FIB) column, showing these performance gains.

While SIMS imaging with a lateral resolution of 20 nm was demonstrated in the late 1980s [1], commercial SIMS instrument still only offer lateral resolutions in the tens of nanometer range (e.g. 50 nm on the Cameca NanoSIMS). In general, the lateral resolution in the microprobe mode is limited by (i) the ion probe size, which depends on the brightness of the primary ion source, the quality of the optics of the primary ion column and the extraction fields in the near sample region (ii) the signal to noise ratio as small voxels generate very few ions, and (iii) the physical dimensions of the collision cascade. In an attempt to further optimize the lateral resolution in SIMS imaging, we have tackled these parameters one by one.

Recently the ORION NanoFab has become the de-facto standard for high resolution FIB applications [2]. This instrument is based on a Gas Field Ionisation Source (GFIS) that can be operated with either He⁺ or Ne⁺. The source has a brightness roughly two orders of magnitude higher than that of the Ga⁺ Liquid Metal Ion Source as used in [1]. As the GFIS provides probe sizes smaller than 0.5 nm, it offers the tantalizing prospect of performing SIMS at resolutions limited only by the collision cascade, which has dimensions of 5-10 nm for He and Ne irradiation at beam energies in the 10-30 keV range.

However, the use of light and non-reactive species (He and Ne) on the NanoFab leads to intrinsically low sputter yields and ionization yields. The lower sputter yields are more than compensated by the significantly higher current densities used in the NanoFab. Ionization yields are addressed by the use of oxygen or caesium flooding. We have shown that order of magnitude increases in the yields for He and Ne are possible using such reactive gas flooding, leading to detection limits of 10^-3 to 10^-6 for lateral resolutions of 10 nm and 100 nm respectively.

Having established the feasibility of performing SIMS on the NanoFab from a fundamental point of view, we addressed the instrumental aspects, focusing in particular on extraction optics providing a high extraction efficiency while minimizing the effect on the primary ion beam, coupled to an in-house developed double focusing magnetic sector spectrometer.

Our results indicate that sub 20 nm SIMS imaging will be possible on this instrument. We will present the latest results obtained on the instrument along with the prospects for combining 10 nm SIMS images with sub nanometer secondary electron images.

References

[1] M. Chabala, R. Levi-Setti and Y.L. Wang, Appl. Surf. Sci. 32 10 (1988)

[2] L. Scipioni, C. A. Sanford, J. Notte, B. Thompson, and Sh. McVey, J. Vac. Sci. Technol. B 27, 3250 (2009)

The cellular plasma membrane is a selectively permeable lipid bilayer that separates each cell from its surroundings. In mammalian cells, the plasma membrane contains numerous different types of lipids, cholesterol, and a variety of different proteins. The distribution of one type of lipid, the sphingolipids, within cellular membranes is of much interest because it can be metabolized to bioactive signaling molecules that modulate many cellular processes. However, until recently, little was known about the sphingolipid distribution within the plasma membrane and the mechanisms that regulate it due to the difficulty of imaging lipids without using bulky labels that may alter their native distribution (i.e., fluorophores). To address this issue, we have used high-resolution SIMS, performed with a Cameca NanoSIMS 50, to directly image metabolically incorporated, stable isotope-labeled lipids in actual cell membranes. The rare stable isotopes, ¹⁵N and ¹⁸O, are first selectively incorporated into the cellular sphingolipids and cholesterol, respectively, by culturing the fibroblast cells in the presence of ¹⁵N-sphingolipid precursors and ¹⁸O-cholesterol. Then the cells are chemically fixed, and coated with a thin iridium layer that increases secondary ion yields during SIMS analysis. Finally, a Cameca NanoSIMS 50 instrument is used to map the lipid-specific isotope enrichments in the cellular membranes with ~90-nm-lateral resolution. This approach revealed that the metabolically incorporated ¹⁵N-sphingolipids are enriched within distinct domains in the plasma membranes of fibroblast cells. In contrast, the ¹⁸O-cholesterol is fairly evenly distributed within the plasma membrane, and is not enriched in the sphingolipid domains. Use of this approach to evaluate the mechanisms that hypothetically regulate sphingolipid organization in the plasma membrane will be discussed. Recent efforts to image the lipid abundances within intracellular membranes will also be presented.