Photoresist thickness measurement is easy with MProbe Vis system.Practically, any photoresist can be measured quickly and reliably: baked or partially baked, thin or thick. Still, the first time measurement may be confusing – there is a wide variety of photoresist types and processing conditions. This application note seeks to clarify the measurement process using examples of baked, and two partially baked photoresist samples. The photoresist measurements are done using 500nm -1000nm wavelengh range to minimize the effect of the absorption. 

 Step 1. Determine  the optical constants of the photoresist.

• Many photoresists are already defined in the material library. Nothing needs to be done in this case
•  Photoresist  specification sheets have information about dispersion parameters. They are listed in the form of Cauchy coefficients – it is 3 coefficients that allow to determine the refractive index (R.I.) of the material at different wavelengths of light.  New Cauchy material can be created and assigned the Cauchy coefficients of the specific photoresist
• If Cauchy coefficients of the photoresist are not available –  use the Cauchy coeffcients of the simiar photoresist, as a starting point.

Step 2. Create the filmstack

Filmstack represents a model of the physical sample – it defines the substrate and material layers. If 3000nm of photoresist is deposited on the Si wafer – filmstack will be Si substrate/ 3000nm PR.  Here PR will be the photoresit material that was defined in the Step 1. 

Step 3. Take the measurement

Measurement is, actually, a two-step process: data acquisition and data analysis. They are transparently handled by the TFCompanion software. During the first measurement one may or may not get a perfect result – filmstack needs to be adjusted. If the thickness of the photoresist is thick enough(> 1um), one can start with the thick  film (FFT based) algorithm. Once the thickness is determined, filmstack can be fine tuned using curve fitting (Marquardt-Levenberg algorithm). At this point, approach is slightly different depending on the processing/conditions of the photoresist(fully baked, not baked, partially baked, etc) If the photoresist is fully baked and the Cauchy coefficients are correct – nothing needs to be done.   Otherwise, optical constants need to be adjusted.

Conclusions

The results show that one needs to be especially careful with under baked photoresist where optical costants are different from the specification.  Correct optical constants can be determined or confirmed by fitting of the model to measured data. Alternatively, optical constants can be measured independently (using spectroscopic ellipsometry, etc.).  Once optical constants are determined – FFT algorithm can be used for further measurement. In the case of thick films, the thickness results of FFT closely match the thickness measured with the curve fitting algorithm. However, there maybe small descrepancies  (<0.1%) when the photoresist has a noticeable absorption. FFT will, typically, give a slightly higher thickness value because it does not use absorption information. The extinction coefficients are very small (<0.01) in the 500-1000nm range but they effect optical measurement in the thick films. At the same time, FFT offers fast and foolproof measurement method, well suited for production environment.