[Zurück]

J. Laimer, C. Forsich, J. Humlicek, H. Störi:
"Closed Loop Control of a Plasma Nitriding Process by Spectroscopic Ellipsometry";
Vortrag: 4th Workshop Ellipsometry, Berlin/D; 20.02.2006; in: "4th Workshop Ellipsometry, Bundesanstalt für Materialforschung und -prüfung", (2006), 20 S.

1 Introduction
Nitriding is a well-established process for the surface hardening of steel parts. On the nitrided part a compound layer, containing iron nitrides is distinguished from a diffusion layer, containing only interstitial nitrogen. Several nitriding technologies are in general industrial use. The plasma nitriding process can be tightly controlled by modifying the nitrogen partial pressure, the plasma power and the temperature. This gives the opportunity to generate a pure diffusion layer without a compound layer as needed for several applications. As the progress of the nitriding process depends on several hard to control parameters, linked to the history of the workpiece, an on-line monitoring tool is desirable. We are proposing to use spectroscopic ellipsometry (SE) as an on-line monitoring tool. Being an optical method, SE is not affected by high temperatures, process gases, plasmas, etc. SE can be used as a monitoring tool or a sensor for closed loop control of processes.
2 Experimental
A spectroscopic ellipsometer (SENTECH, Berlin, model SE801) was attached to a prototype scale plasma nitriding and CVD plant (custom built by RÜBIG, Wels/Austria). Samples were discs made from chromium alloyed steel with a thickness of 5 mm and a diameter of several cm. The sample surfaces were polished to mirror like condition prior to the deposition or nitriding experiment. For experiments on closed loop control ellipsometric spectra are fitted on line and the fitted parameters are used as input into a software controller, influencing a parameter of the CVD or nitriding process, e.g. the nitrogen flow. A Drude-Lorentz model of the optical response was fitted to the raw ellipsometric data. This results in position, width and damping of individual oscillators. Typically 3 oscillators were used, whereby one has zero frequency, modeling the contribution of free charge carriers. For the continuous fitting during the experiments, an effective interface model was applied.
3 Results and discussion
Observing the time evolution of a measured quantity, e.g. the ellipsometric angle Y at a photon energy of 2,75 eV during a nitriding run, the effects of heating and plasma cleaning are clearly discernible, as expected, given the monolayer sensitivity of ellipsometry. After the onset of the nitriding process a plateau is quickly reached, where a diffusion layer is formed. Once the formation of the compound layer, i.e. the formation of iron nitride, starts, a rapid change of the ellipsometric signal is observed. The effects are even more clear using the position of the VIS-UV Lorentz oscillator as the quantity observed over time. Off-line investigations using atomic force microscopy and x-ray diffraction suggest, that at this time really the formation of a compound layer starts. If the process is interrupted shortly after the onset of the compound layer, the iron nitride dissolves during the cool-down process step. The only trace left from the phase transitions is slight roughening of the surface.
Closed loop control, using the output of the fitting process, has been demonstrated with a number of different control algorithms. The closed loop control system permits nitriding very close to the transition to the compound layer formation.

Acknowledgements: The authors are grateful for the support of this project by the European Commission within the fifth framework program (CRAFT Project No. CRAF-1999-71740, contract No. G5ST-CT-2002-50362) and for the intense cooperation with SENTECH Instruments in Berlin and with RÜBIG Anlagentechnik in Wels, Austria.