FLIG® 5

The requirement for high depth resolution dynamic SIMS arises from the reduction in device size in the semiconductor industry. With the use of low implantation energies and new technology dependent on delta-doping and sharp interfaces, it is increasingly important to have access to depth profiling techniques which can quantify these structures both accurately and reproducibly.

When profiling with energetic oxygen beams, the kinetic energy of the incident particles is transferred to the near surface region of the sample, creating an altered layer in which atomic mixing has occurred. The depth of this layer is approximately 4nm per keV for an O2 beam, and this imposes an absolute limit on depth resolution. Hence, it is necessary to use energies well below 1keV for profiling of shallow junctions.

Another important factor in the analysis of shallow implants is the transient region which occurs at the surface and at matrix interfaces. At the beginning of a shallow profile, the ion and sputter yields vary rapidly as probe atoms are incorporated into the analysed surface, and the surface chemistry changes. Similar effects occur at matrix interfaces. Whilst this behaviour persistes, the profile depth profile is not quantifiable, and any features lying within the transient will be distorted. The thickness of the region, and hence, the amount of lost information, can be reduced by using low impact energies.

In a conventional ion gun, ions are transported through almost the whole ion-optical column at an energy determined by the anode voltage. Thus, to attain a 250 eV impact energy (on a grounded sample) the anode must be set to 250 V and the beam travels through the column at this energy. At such low energy, space charge effects and aberrations of the wide beam seriously limit the final intensity of the probe and impair the probe shape.

In the floating ion gun, almost all of the column is floated to a negative potential and the beam is accelerated to a more viable transport energy between the extraction region and the final lens. Inside the final lens, the beam is decelerated to the desired impact energy. Thus, for a 250 eV impact energy, the anode is set to 250 V and the float could be -3 kV giving a transport energy of 3.25 keV. This provides a signicant reduction in beam aberrations. In the FLIG, the Wien filter electrostatic plates and alignment units (including a bend to reject neutrals) are all referenced to the float voltage.

Shallow junction profiling requires the use of a low energy primary beam in order to minimise the effects of atomic mixing induced by the beam. As sputter yield reduces with lower energies, it is vital that the low energy probe has a high current density. Figure 1 shows characteristic plots of probe size versus beam current for the FLIG 5.

The FLIG 5’s high brightness duoplasmatron source and floating column optics deliver exceptional probe inten-sity, in comparison with conventional systems, facili-tating low energy profiling with acceptable erosion rates.

To attain high depth resolution without sacrificing erosion rate, the bottom of the analysis crater must re-main flat through the profile. A good probe shape, with minimum aberration tails, is essential to minimise cur-vature at the sides of the crater. Reducing the extent of this curvature enhances depth resolution and dynamic range, as well as allowing the use of smaller scan fields and hence shorter analysis time.

Figure 1. Current vs spot size for the FLIG^® 5.

The FLIG’s floating column transports the beam through most of the optics at high energy (generally between 2.5keV and 5keV). This greatly assists in reducing beam spreading in the column and concomi-tant aberrations in the probe. The result is sub-nanometer depth resolution at low energy, with profiles showing high dynamic range at all energies.

The depth resolution capability is demonstrated in Figure 2, which shows profiles of a Si-Ge superlattice. Grown by MBE, this structure has alternating 1nm layers of Si and Ge. The low energy profile shows a 45% valley between Ge peaks 14 and 15, showing the feature to be easily resolved.

Figure 2. Depth profiles of SiGe lattice

A remarkable feature is the apparent increase in resolution with depth in the low energy profile. A cross-sectional TEM image of the sample revealed that the upper layers were buckled, causing the lower resolution of the top layers.

Control of the voltage settings in the FLIG is greatly sim-plified by the use of a computer software interface. This allows many useful features to be built into soft-ware, the most valuable being the facility to save complete sets of operating voltages. Critical control voltages such as extraction and alignment are refer-enced to other supplies rather than ground to simplify tuning of the column.

The system is ready for use with automated systems which can command a change of preset conditions, when required, using ascii commands.