Sample Measurement with GDOES

Ideal samples for Pulsed RF GDOES are flat, rigid and with a diameter between 1.2 cm and 25 cm. But most solid samples can be measured, possibly after some preparation. Non-flat samples, flexible or porous ones will require the implementation of adequate strategies.

7. Hygroscopic or flammable samples can also be done with care. An accessory (“Li bell”) was designed to transport such samples to the instrument under inert atmosphere and running the analysis without contact to air.

8. Radioactive materials require secure confinement. A special instrument dedicated to the challenging measurement of C and N (VUV elements) in radioactive Pu oxides (non-conductive) was developed in France.

Pulsed RF GDOES is applicable to “thin” and “thick” films to obtain elemental depth profiling and will show in minutes:

• Which elements are present in the sample
• At which concentration levels
• Whether the sample is homogeneous in depth
• If any coatings or surface treatments were applied to the sample
• How thick are the coatings
• Whether there was any contamination at an interface
• Whether there was any oxidation of the sample
• Whether there was any diffusion in the layers

This was not possible until HORIBA introduced “DiP” in 2016. Of course quantitative results (cc vs depth) are often presented, but the depths and layer thicknesses always resulted from calculations – relying at some steps on the use of profilometers to accurately provide the mandatory depth information.

Changes in density specifically, were not possible to estimate through the standard GD calculations affecting the accuracy of depth calculations, in the case of PVD coatings.

HORIBA has built in an interferometer within the GD source – this is the patented “DiP” device. DiP allows the direct measurement of erosion rates  and layer thickness in parallel to the GD measurements. For the first time, GD and DiP can measure erosion rates.

The range of applications is vast as thin and thick films, conductive and non-conductive, can be easily measured with Pulsed RF GDOES.

A strict division between thick and thin films is uncertain since the difference lies mostly in application rather than in film thickness itself. Various definitions are encountered in the literature and most of them originated from electronics or microelectronics-related applications. This is not surprising because the process of fabrication of electronic components involves many different approaches for creations of films and layers (e.g., plasma, electroless or electrochemical deposition, epitaxy, ion implantation, sputtering, thermal oxidation). Further application of lithography will build up transistors, dielectrics, metallization, and other functional elements.

A widespread definition, however, is that a “thin film” is a microscopically thin layer of material deposited onto a metal, ceramic, semiconductor, or plastic base and is typically less than 1 μm thick.

A practical “definition” could also be derived from the capabilities of the instrumentation in use. When the material to be analyzed is sputtered, the practical limit is the thickness range achievable in a reasonable amount of time. This practical limit is in the micrometer range for SIMS, but Pulsed RF GDOES is perfectly capable of sputtering rapidly down to 100-150 microns – sometimes even deeper.

At the same time, Pulsed RF GDOES can measure thin films (from 1 nanometerer upward) as optical signals can be recorded much faster than the erosion of the material if suitable instrumentation is available.

So within a single experiment, Pulsed RF GDOES can offer information on top layers and deep interfaces. The following example shows results obtained on a hard disk.

There is no absolute answer to this question. Nanometer depth resolution has been shown, and results on atomic monolayers have been published. However, operating conditions that influence not only the crater shape, but also the sample itself will play a crucial role.

The illustration below is reproduced from the ISO standard on Zn coatings. The thickness of interface can be estimated by looking at the location of 84% and 16% of the Zn signal, and showing how thick this interface can be on rough industrial samples. This is the reason why people often consider coating weight integrating the total Zn amount.

In standard operating conditions, the plasma covers the entire surface of the sample facing the anode and it is usually relatively straightforward to find operating conditions leading to a flat crater required for achieving high depth resolution.

However, if the conditions are not optimized, the crater shape can be concave or convex and the depth resolution is correspondingly degraded.

Sample roughness is most often the crucial parameter to consider when one looks at ultimate depth resolution (nm scale). Let us imagine two samples with similar flat top surfaces. The first one was obtained by deposition on a flat substrate, and the second one was obtained on a rough substrate as shown in the TEM image below.

As GD has no lateral resolution and averages signals from the entire sputtered surface, it is easy to understand that, on the rough substrate, material from the substrate will start to be sputtered while some top material is still present, and the interface will therefore appear larger than in “reality”.

In some cases where the sample features a surface waviness, the use of a smaller anode diameter may possibly help to minimize the effect of the variations on the observed results.

This question is often asked by surface scientists who doubt that a fast sputtering instrument can be used for surface analysis.

The real answer is that the source is immediately stable after starting, and that optical signals can be recorded much faster than the erosion.

To go even further, in SIMS the depth resolution is improved by lessening the energy on the incident beam and rotating the sample. We do not rotate the samples in GDOES, due to the fast sputtering rate of the instrument. And the energy of the incident particles is already very low.

Simply reducing the applied power does not improve the depth resolution: the two are not correlated – but it decreases the signals available which is not what we wanted.

Conversely, using pulsed operation does improve the depth resolution witho ut necessarily reducing the Sputtering Rate (SR), as it is possible to use higher instantaneous powerwith variable duty cycle. The reason for depth resolution improvement is probably due to the minimization of the redeposition.

When sputtering takes place, most of the sputtered material enters the plasma but a part will redeposit on the sample surface. When the plasma is stopped, as the Argon flow is maintained, this redeposited material is flushed away and does not contribute to the signals in the next pulse.

Many publications show interesting results on layers thinner than 100 nm, 10 nm or even at the nm level. The example below is the analysis of a mirror for X ray featuring 60 stacks of 3 layers, the thinnest one being 0.3 nm ! An almost constant depth resolution at nm level was achieved on this ideal sample.

The sputtering efficiency is usually expressed in mass/time and indicates the sputtered mass removed from a material per unit of time. The erosion  ate simply reflects the depth/time. If the density of the material is known (or calculated), one can be calculated from the other.

They depend on the operating conditions, as well as the actual material. For example, sputtering will be more rapid with a more energetic plasma. they are also material dependant. In same operating conditions, a Zn sample will sputter faster than a Cu sample. It should never be forgotten that a measurement results from the interaction between the plasma and the sample, and that the understanding of the interaction mechanisms is therefore crucial.

The following example illustrates the changes of SR between 2 layers. The sample is an organic rubber on stainless steel. The rubber sputters less rapidly than the steel. When the interface is reached, the material facing the sample is a mixed stainless/rubber and the sputtering rate steadily increases. Hence the rest of the rubber is sputtered more rapidly. This is why the C signal peaks: it does not correspond to an increase of concentration, but to an increase in SR.

Quantification takes the SR changes into accounts. One can also divide the raw qualitative signals by the Fi signal (Total Light always measured in the instruments) that give an estimate of SR changes.

Both quantitative analysis and qualitative analysis can be obtained by GDOES.

GD measures intensities, and as with all comparative techniques, requires calibration to provide concentrations vs. depth (named “QDP” Quantitative Depth Profile or “CDP” Compositional Depth Profile) from the measured Intensities vs. Time (Qualitative Depth Profile).

Calibrating the GD for a dedicated application is not always easy mainly due to the absence of relevant reference samples. With pulsed RF however, more samples can be used for calibrations, as opposed to just the bulk metallic ones usually considered at the origin of GD. This offers a lot of flexibility. Samples can now be bulk (conductive or non) or layered samples – conductive or non. Oxide layers can be mixed within the calibrations to provide data points for O: polymer layers can be added when organic layers are of interest. In addition, if at least one customer sample known in thickness and composition is available, it can be used directly for the calibration. The introduction of DiP providing the measurement of the depths is also greatly changing and simplifying the calibration procedures.

There is no simple answer to this question. A presentation on precious metals – available on request - was made at the 5th GD day showing ppm or sub ppm data for most elements.

Of course the Detection Limit of an element is dependent upon the sensitivity of the emission line selected for this element, as with any emission spectrometer. It also obviously depends on the performance of the optical system.

But in GD, the detection limits also depend on the material and the operating conditions. One should always keep in mind that for GD, the sputtering and the excitation are separate processes.

If the same material is measured with an applied power of 60 W compared to 30 W, the Sputtering Rate is changed by a factor of 2 and twice the amount of material is sent to the plasma, leading approximately to double intensities and clearly better DLs.

The size of the anode also plays a role if the optical system can accommodate various diameters: obviously a 7 or 8 mm anode will provide more light than 4 or 2 mm ones (but the crater shape will not be as flat).

In the example on precious metals, the instrument was a Profiler HR, the anode diameter was 8 mm and the power was set over 100 W.

However, it is not always possible to increase the power for all materials: low melting point alloys or fragile samples require soft conditions to be applied. In addition, since the main interest of Pulsed RF GDOES is its capability for depth profile analysis, emphasis is usually put on depth resolution (flatness of the crater shape most often requires soft conditions to be applied.)

This might not be necessarily detrimental to the ability to detect traces if one considers that even if an element is a trace in the overall composition of a layer (possibly far below the Detection Limit of GD), in case it migrates due to the process and accumulates locally, it could be easily detected. Traces of Pb at the surface of ultra pure Cu samples have, for instance, been measured.

By the same token, GD is ideally suited to detect contaminants on interfaces that could even be monolayer. The example below shows a SiO₂/Si featuring (in blue) a residual Ca contamination on an interface, nicely resolved and seen.

In average operating conditions, metals are sputtered at a rate of 1-5 μm/minute. A 100 nm layer could therefore be sputtered in 3 s – 10/15 s in pulsed mode. A thermal treatment on steel in which elements diffuse down to 50 μm could be checked in 12 min.

With UFS, a polymer film can also be rapidly sputtered.

Such fast sputtering requires a fast detection system, even very fast for thin films analysis to be able to adequately follow the varying signals.

It is very easy to test samples because sample does not need to be inserted into a UHV chamber.

In addition, large samples can be easily measured. However, this simplicity still requires that users carefully handle the sample (to avoid contaminations) notably if surface analysis is needed. The ISO procedures developed for UHV techniques can be applied with benefit to GD.

Only the GD source and the MgF₂ lens need regular maintenance. This is routine operation and takes only a few minutes weekly.