Comparison with Other Techniques: Surface Analysis

It is very important to realize the uniqueness of Pulsed RF GDOES for surface and depth profile analysis of materials when compared to the principal “Surface Techniques”. Most surprising for surface scientists is that Pulsed RF GDOES is fast, and does not require an Ultra High Vacuum (UHV).

There are many different and complementary techniques currently used in the field of surface and thin film analysis. They involve bombarding the sample with incoming (incident) particles and monitoring the ejected particles. The precise method being employed is differentiated from the others according to the identity of the respective particles.

Most of these methods require the analysis to be performed in an ultra-high vacuum apparatus. The figure below illustrates the basic principles and the table lists the most common techniques and their acronyms.

In each of the methods wherein the incident particle is either an electron or an ion, measures must be taken to insure that the sample surface is electrically conductive. Thus, for insulating materials and films such as oxides, glasses, and polymers, the experiments are not straightforward (charge compensation is needed). Pulsed RF GDOES does not need any charge compensation.

The fact that the instruments are maintained at high vacuum levels and have very sophisticated components makes them expensive and virtually requires that the operator have a Ph.D.

On the other hand Pulsed RF GDOES is very easy to use and is comparatively inexpensive.

The use of the above methods in depth profiling applications, rather than surface analysis is dependent on the sampling depth of the incident and ejected particles. For XPS and AES, this value is approximately 3 monolayers (≈10 angstroms), SIMS and SNMS, 10 monolayers, and on the order of 100 monolayers for RBS. Subsurface information can only be obtained beyond these levels through the removal of sample material usually through sputtering with a high energy ion beam. This step is implicit in the SIMS/SNMS experiments, but is usually included as an auxiliary feature in the other instruments. In XPS, for example, one would alternate between bombardment of the sample with ions and X-rays to generate a depth profile.

Pulsed RF GDOES, combines the sputtering and excitation with a single plasma. Sputtering is also comparatively much faster and much more delicate than the techniques used in SIMS or XPS as incident particles have a low energy.

One of the most compelling reasons for using these surface analysis methods is the ability to generate elemental maps of the sample surface. These images are effectively similar to scanning electron micrographs, but with element-specific information. In the cases where electrons are the incident particles, spatial resolution on the order of 5 nm can be achieved.

Pulsed RF GDOES on the other hand has no lateral resolution as signals are averaged over the sputtered area (several mm in diameter).

Rather than comparing GD to these techniques it is better to emphasize their complementarity as recently proved in publications where XPS measurements were done within GD craters.

The GD plasma has a double role. First it sputters the material and then it excites the sputtered species.

Essential to GD operation is the physical (spatial) separation of the sputtering (at the sample surface) and excitation mechanisms that take place in the plasma. Simply, one could say that sputtered atoms, when they enter the gas phase, have “forgotten” where they were coming from.

Matrix Effects are therefore greatly reduced in GD operation unlike Spark or SIMS.

This important feature greatly simplifies the calibration of the instrument as samples from different matrices can be calibrated together only by taking into account their relative sputtering rates.

GD has much lower Matrix Effects than SIMS or Spark, as the excitation and sputtering are separated.

Matrix Effects are well known in SIMS for surface analysis and in Spark Emission for bulk analysis.

By definition, Matrix Effects are a "change in the intensities or spectral information per atom of the analyte arising from a change in the chemical or physical environment, i.e. they are changes in elemental intensity, not simply related to changes in composition of the element".

This being said, matrices have an influence as operating conditions must be adapted to the type of samples to be measured.

SIMS relies on high energy (2-5 keV) bombardment (sputtering) of the sample using an ion beam (gun) in high vacuum (<10^-7 Torr), and mass analysis of those species which are ejected as charged species (ions).

SIMS is very sensitive to surface condition: the presence of oxides for instance, drastically enhances the secondary ion emission mechanisms.

Pulsed RF GDOES is similar to SIMS in that sputtering is the means of removing material from the sample surface. In contrast, though, the glow discharge is a reduced-pressure (a few Torr) plasma that generates the sputtering ions in situ from a low flow of argon. These ions are attracted to the sample (cathode), arriving with kinetic energies of ~50 eV. The respective ion currents, in the microampere range for SIMS and ~1 ampere for the Pulsed RF-GDOES, result in much greater ablation rates for the latter, μm/min vs. nm/min. It is also much more matrix independent.

On the other hand, the analyte detection efficiency for SIMS is orders of magnitude higher than for the photon production/collection in Pulsed RF GDOES. Thus, very high absolute sensitivities may be achieved with SIMS. The combination of slow erosion and high detection efficiency results in SIMS limits of detection (LOD) that are usually expressed in units of atoms/cm³ or monolayers, while the Pulsed RF GDOES LODs are expressed in terms of ppm. On a weight percent basis, SIMS detection limits are lower and in the order of ppb-ppm across the periodic table.

Published data have, however, shown the progress in GD over the years. In practice, SIMS users will select GD for the applications where GD will perform adequately and where its speed and ease of use will be of great benefit, keeping the SIMS machine available for other challenging tasks where its unique capabilities will be most beneficial.

There are interesting similarities between SNMS and Pulsed RF GDOES in the sense that in both cases, the erosion is separated from the excitation/ionization mechanisms, and that the information is coming from the neutral species eroded. A presentation at a GD Day provides comparative information.

In Japan, for many years, 65% of the GD users are also XPS users and the two techniques are often used complementarily.

XPS (also named ESCA) is the most popular surface technique with thousands of published applications. A surface irradiated with an X-ray light source emits electrons (photoelectric effect) that are measured. The strength of the technique is to gain information about the chemical environment of the surfaces.

XPS measures all elements except H. Depth profiles are done by coupling a sputtering ion gun with the XPS: each time a layer is sputtered, the XPS will look at what remains on the surface (where in GD the sputtered material is the analyte).

Sputtering in XPS is slow and the maximal depth achievable in practice is about 500 nm. When one wants to look at an embedded interface, GD with its fast erosion speed can be used: the GD discharge will be stopped just before the interface is reached (easy to do as GD signals are displayed in real time) leaving this interface unaltered, and the sample can be further introduced in the XPS chamber.

GD and SEM were often used in parallel to characterize materials.

More recently Professor Ken Shimizu from Japan has published a book (that is now a best seller for this type of work with over 6000 copies sold) showing that recent advances in Field Emission SEM could be used to obtain valuable topographic details on surfaces if these surfaces are adequately prepared, and he has shown that RF GD Sputtering could play a key role.

“This arises from its unique ability to sputter both conductive and non-conductive surfaces with Ar⁺ ions of very low energies, less than 50 eV, and high current densities of ~100 mA cm^-2. Based on the typical Ar⁺ ion penetration depth (e.g., 0.1 nm at 100 eV for Cu [3]), it appears that altered layer formation is insignificant here. In addition the high current density of Ar⁺ ions ensure sputtering to proceed at very high rates, namely about 1- 10μm min-1, making total sample treatment time extremely short, less than 1 min including sputtering for normally less than 10s. Further, RF GD sputtering creates sharp steps along the boundaries of different materials due to the so-called differential sputtering effect, with the height and sharpness of steps being controlled precisely just by changing sputtering time for a given RF power and Ar pressure”.

Raman can, of course, be considered as a surface technique, since the collected signals come from the interaction area with the laser. In some cases depth profile information can be derived by using different wavelengths, as in the example below (strained Si by implantation of a SiGe layer).

GD craters can also be done on a material, and Raman signals collected within the sputtered area provide molecular information of an embedded layer, as in the example below studied within a Ph.D. thesis on Cr speciation.

ICP has more surprisingly also been used for surface analysis. This application was presented at the 5th GD day. In this case, a flow of acid was spread onto a surface and the dissolution products were measured in real time (on a simultaneous instrument equipped with the GD software).

Obviously Spectroscopic Ellipsometry is the technique that comes to mind when people think about thin films in HORIBA Instruments. The obtained information with the 2 techniques are different (chemical depth profile in the case of Pulsed RF GDOES and thickness with DiP, optical constants, gradients, thickness in the case of SE) but complementary as DiP is ideal for reflective layers where SE requires layers to be transparent.

A lot of work is done cooperatively, with GD results being used to establish the models for SE, in the case of new materials, and some SE algorithms being implemented within the DiP software in the case of analysis of simple transparent layers for example.