Ag on glass: Our first deposition experiment
Using the new DD1 sputtering device we deposit Ag on microscope slides. 10 samples were prepared: With constant electric power we varied the deposition time:
Sample Ag sputtering time [s]
dd1_ag_ms_08_06_2004_0 0
dd1_ag_ms_08_06_2004_1 8
dd1_ag_ms_08_06_2004_2 10
dd1_ag_ms_08_06_2004_3 12
dd1_ag_ms_08_06_2004_4 14
dd1_ag_ms_08_06_2004_5 16
dd1_ag_ms_08_06_2004_6 18
dd1_ag_ms_08_06_2004_7 20
dd1_ag_ms_08_06_2004_8 22
dd1_ag_ms_08_06_2004_9 24
dd1_ag_ms_08_06_2004_10 26
The reflectance and transmittance spectra of these samples have been measured with the S2 spectrometer system and stored to the database.
To get an overview, the measured spectra are displayed using the Collect tool which is delivered with CODE. Here are the transmission data
and here the reflectance spectra:
We analyze the spectra with a CODE configuration that computes R and T in the spectral range of the S2 spectrometer. First of all we check if the substrate spectra are consistent with the expectations based on the optical constants of microscope slide glass from our database. That is the case as the following comparison shows:
We now load the spectra of sample #5. For the optical constants of Ag we have nothing in our database yet. We start with data taken from the original CODE database. There are several versions - we take the object called 'Ag model' because this will allow us to change the model later on if necessary. Fitting the thickness of the Ag layer in the model the following best fit is achieved:
The agreement is not bad although there is a significant difference between model and measurement. There can be several reasons for this: The measurements could be wrong, the layer could have a surface roughness, there could be a depth gradient in the Ag properties within the layer, or the Ag sputtered with our device is different from the one that was used to determine the optical constants found in literature. Most likely is the last reason: Ag obtained by sputtering may have more internal damage (grain boundaries, impurities) than material evaporated under very clean conditions. Grain boundaries or impurities will increase the scattering of the electrons, i.e. reduce the conductivity of the metal. In the framework of the simple Drude model (that is used in our Ag model) this means that the damping constant should be higher. Introducing the damping constant as additional fit parameter, we give CODE the chance to take into account different Ag qualities. The agreement of model and measurement is almost perfect now:
Being satisfied with the current model we try if we can fit all measured spectra using the method. We process all samples making use of CODE's batch operation feature. Because there are only two fit parameters the analysis of all sample spectra takes a few seconds only.
A second view is created which summarizes the results from the batch control window:
The fits are all very good (the fit deviation is below 0.00001). The thickness shows a nice linear relation to the deposition time. The deposition rate is about 0.6 nm/s. In all cases the electron damping is close to 430 1/cm. This value is significantly higher than the 145 1/cm from the 'literature model'.
We can conclude that we should store our own Ag version (with a damping constant of 430 1/cm) to the VirCoC database. We name it 'Ag model (DD1)' to indicate the machine which is used to produce this kind of Ag.