I have to support Marinus Hopstaken in this. Timo Sajavaara is exactly right (and very reliable). Various methods have been suggested, sometimes with excessive confidence in my view. Taking these in turn:

LIBS (Laser-induced breakdown spectroscopy: a sort of GDOES with laser ablation). This will work but like GDOES (glow discharge optical emission spectrometry) is destructive and needs sample-matched standards.

EDX (meaning energy dispersive X-ray spectrometry on the SEM). No depth scale, so very hard to quantify.

TEM. Transmission electron microscopy is of course a very powerful technique, but in this context it has various problems: SAD (selected area diffraction) is unsurpassed for identification but hard to use for quantification; EDX can be used but has the same problems as on the SEM; sample preparation is hard; destructive. EELS (electron energy loss spectrometry) can indeed be used and may work well, but also may not be sensitive enough. Another problem of TEM is that the field of view is very small so that the question of whether the result is representative is hard to answer.

SIMS (secondary ion mass spectrometry). Needs sample-matched standards and even then quantification is always non-trivial because of the sensitivity of the ionisation yield to secondary (or tertiary) effects. Destructive. The particular problem of SIMS in this application is that the material is not homogeneous, and there are a range of very tricky SIMS artefacts that attend the sputtering through interfaces.

ICP-MS (inductively-coupled plasma mass spectrometry). Exceptional (deep sub-ppb) sensitivity, but of course destructive and no depth resolution. For this application you would need LA-ICP-MS (laser ablation ICP-MS) or something similar for depth sensitivity, and you still have the quantification problem.

The optical absorption techniques are the most straightforward (provided the samples are transparent), and are probably very sensitive, but intrinsically have no depth resolution so that the total number but not the density of optical centres will be detected.

RBS (Rutherford backscattering spectrometry) on the other hand has deep sub-monolayer sensitivity for Ag in SiO2 and is a standard-less technique capable of 1% absolute accuracy in determining quantity of material (see Analytical Chemistry 84, 2012, 6061). By the way, you also get the SiO2 thickness, and if you turn on the X-ray detector you also get all the impurities in the oxide. So with the RBS you also get EDX, but at very much better sensitivity than on the SEM (no primary Bremsstrahlung!). RBS is not destructive and needs no sample preparation.

You should be able to get RBS at K.U.Leuven through the METACEL consortium that Gent is part of.

Techniques are:

1. Solid state ICP-AES

2. Solid state UV-vis if initial amount of Ag is known and Ag NPs are coloured.

3. Quantitative EDX analysis against a known sample.

First one is the easiest one. If solid state ICP-AES is not available, dissolve the film in aqua regia and check by solution ICP.

You can also use electron energy loss spectroscopy (EELS) in a TEM or STEM to quantify the composition of Ag in your film. With this technique you could also measure the size of the nanoparticles.

Good luck!

Thanks for the excellent suggestions. Is there any detail information about the solid state ICP-AES and solid state UV-vis? Thanks..

As it has been stated before, I'd recommend SEM. Not sure how TEM would work unless your sample can be put on a TEM grid. As already stated, this will allow you to quantify coverage directly, as well as perform sizing (although I'd recommend sizing with UVVis, DLS or other methods as intrinsic thresholding errors are inevitable with SEM).

In addition to the other techniques mentioned, I'd add SIMS. It's very accurate and compatible with thin films.

A simple and quantitative measurement for this type of samples would be Rutherford Backscattering Spectrometry (RBS) if Ag content is higher than at ppm level. This kind of measurement should be available in Gent as well. If you want quantitative information on all the sample elements, like for instance amount of impurities introduced during the deposition/annealing, you could go for heavy ion elastic recoil detection analysis (HI-ERDA) which belongs to ion beam analysis techniques as well. The beauty of these techniques is that unlike most other techniques, they do not require any reference samples for quantification.

TEM will allow you to image the particles. If you use high-angle annular dark-field imaging ina scanning TEM you can readily distinguish the high-Z Ag particles from the low-Z matrix. Of course you can also do chemical mapping to distinguish the particles. Here, EDX is probably most promising. But you can also use EELS although the Ag M-edge is probably not ideal in terms of sensitivity. Spatial resolution of both techniques are in the nanometer or even Angstrom range, depending on the TEM used.

Of course, TEM requires proper thinning of the sample and you need somebody having some experience in thinning procedures.

SIMS I think it will be better

In my opinion, to get a quantitative overview, SIMS would be the best. I already performed such measurements both for Au and Ag in Si samples, and it gives a good resolution. Additionally, SIMS will give you a depth profile of the contamination in the layers. If SIMS is applied to a bigger area of interest in the sample, it will give better statistics for contamination measurements.

I believe that (HR-)TEM would also be a good technique, but the preparation is much more complex than for SIMS and therefore not really feasible for analyzing larger quantities of samples.

In addition to SIMS (which is really an optimal method for this purpose) I would also recommend GD-OES. I used it once to determine the distribution of Ag nanoparticles after wet-chemical etching of silicon. It is similar to, if not the same as, ICP-AES.

I hate to say this as an SIMS expert, but this is probably the least appropriate technique for this particular materials system if your main objective is to obtain quantitative analysis of Ag for the following reasons:

- explicit need for calibration standard

- inhomogenous phase (Ag nano particles are embedded in oxide matrix) with probably highly different ionization yields. Ag-yields may actually depend on dispersion (i.e. particle size distribution), such that samples with similar Ag-loading (average Ag-content) may give rise to different ion intensities., which obviously prohibits any meaningful quantitative analysis.

SIMS will be able to obtain qualitative depth profile, allowing to assess in-depth uniformity and you may be able to do some relative sample-to-sample comparison, provided particle size distribution does not vary too much.

I agree with previous comment by Timo recommending RBS (providing sufficient Ag above detection limit). For lower impurity levels, ICP-MS / ICP-OES might be most appropriate. TEM would be great for particle size distribution, provided you can sample sufficiently large number of Ag-nano particles, maybe in top-down TEM (after etching window in Si-wafer; depending in oxide thickness).

If I would have to decide I would try, at first, with EDX, although this is not the best (in terms of detection at such low level) technique for such sample. Also you could try LIBS - it will show gold atomic lines if the gold particles are there, although without proper calibration this also would not give any quantitative results.

Amount of Ag ... in SiO2 will also matters the selection of right tool...If it is around few % of your matrix SiO2.

then the simple and effective tool is GIXRD ....as it is the film ... the choice of the glancing angle will be depending on your film thickness.

Beside of this If possible .... go for TEM ...

and take selected area diffraction pattern ....as Ag is in your film ...It will be very different morphology then SiO2 ... diffraction pattern will be different ...You can also do the TEM/EDX and quantify.

I have to support Marinus Hopstaken in this. Timo Sajavaara is exactly right (and very reliable). Various methods have been suggested, sometimes with excessive confidence in my view. Taking these in turn:

LIBS (Laser-induced breakdown spectroscopy: a sort of GDOES with laser ablation). This will work but like GDOES (glow discharge optical emission spectrometry) is destructive and needs sample-matched standards.

EDX (meaning energy dispersive X-ray spectrometry on the SEM). No depth scale, so very hard to quantify.

TEM. Transmission electron microscopy is of course a very powerful technique, but in this context it has various problems: SAD (selected area diffraction) is unsurpassed for identification but hard to use for quantification; EDX can be used but has the same problems as on the SEM; sample preparation is hard; destructive. EELS (electron energy loss spectrometry) can indeed be used and may work well, but also may not be sensitive enough. Another problem of TEM is that the field of view is very small so that the question of whether the result is representative is hard to answer.

SIMS (secondary ion mass spectrometry). Needs sample-matched standards and even then quantification is always non-trivial because of the sensitivity of the ionisation yield to secondary (or tertiary) effects. Destructive. The particular problem of SIMS in this application is that the material is not homogeneous, and there are a range of very tricky SIMS artefacts that attend the sputtering through interfaces.

ICP-MS (inductively-coupled plasma mass spectrometry). Exceptional (deep sub-ppb) sensitivity, but of course destructive and no depth resolution. For this application you would need LA-ICP-MS (laser ablation ICP-MS) or something similar for depth sensitivity, and you still have the quantification problem.

The optical absorption techniques are the most straightforward (provided the samples are transparent), and are probably very sensitive, but intrinsically have no depth resolution so that the total number but not the density of optical centres will be detected.

RBS (Rutherford backscattering spectrometry) on the other hand has deep sub-monolayer sensitivity for Ag in SiO2 and is a standard-less technique capable of 1% absolute accuracy in determining quantity of material (see Analytical Chemistry 84, 2012, 6061). By the way, you also get the SiO2 thickness, and if you turn on the X-ray detector you also get all the impurities in the oxide. So with the RBS you also get EDX, but at very much better sensitivity than on the SEM (no primary Bremsstrahlung!). RBS is not destructive and needs no sample preparation.

You should be able to get RBS at K.U.Leuven through the METACEL consortium that Gent is part of.

Also it is possible to try AES (Auger electron spectroscopy). It gives not only the quantification in atomic%, but also the high resolution chemical images, i.e. the lateral distribution of Ag in the film.

You can try following:

1. from TEM/HRTEM image analysis you can find the particle size and size distribution.

2. Indirectly, if possible, remove the SiO2 layer by selective chemical etching and then collect Ag and get your quantification.

Rutherford Backscattering spectrometry is the best method to quantify the Ag in the thin film with monolayer sensituvity..!!

Chris Jeynes is quite right when he states that if you turn on the X-ray detector in an RBS analysis you will get also SiO2 film contaminationa by EDX, the only missing point here is to put a name on that, and that is PIXE (Particle Induce X-ray Emission), quite a good possibility indeed. This is basicaly EDX similar to SEM-EDX but using protons or alpha particles like RBS, and therefore the continuum X-rays background is nearly a factor of 2000 lower.

Quantification by PIXE in this case should have no problems what so ever, but both L X-rays and K X-rays should be used to identify contaminations from the substrate (assuming there is one).

In extreme cases, where high detection limits are required it is even possible to use high resolution PIXE at some laboratories where there are WDS detectors.

If PIXE detection limits are not enough (namely because the contamination is a levels below 10ppm of the sample mass, or less than 0.01pg absolute), then there is always a possibility to use Sy-XRF (Synchrotron based XRF), which is absolutely non-destructive and has also mapping cappacities in some cases.

KULeuven has RBS, no need to go anywhere!

http://fys.kuleuven.be/iks/nvsf/experimental-facilities/rutherford-backscattering-spectrometry-and-ion-channeling

You can determinate the quantity of silver by Inductively coupled plasma mass spectrometry (ICP-MS), is a type of mass spectrometry which is capable of detecting metals and several non-metals at concentrations as low as one part in 1012 (part per trillion).

Although quite late, I would like to add grazing incidence XRF (TXRF): By using monochromatic incident radiation, you can fully control the penetration depth of the X-rays with nm sensitivity, and you are thus able to quantifiy the concentration of the Ag within the SiO2 matrix by a variation of the incidence angle and the measurement (and fit of course) of the Xi and Ag fluorescence!

Best regards, Dirk