Quantitative Spectral Analysis with AAnalyzer

From peak fitting to the layer model: XPS, Raman, and luminescence

 

What this workshop is about

 

Quantitative interpretation of a spectrum rests on decomposing it into physically meaningful components. In practice that peak fitting is one of the main bottlenecks in quantitative analysis: it requires expert judgment to decide how many components are present, what line shapes they have, what background is required, and how their parameters should be constrained. The result tends to be subjective and hard to reproduce.

 

This is a hands-on introduction to a set of free tools built to address that problem directly. The centerpiece is AAnalyzer, which fits the raw data under the Active Background Approach rather than subtracting a background before fitting. Around it we will cover AAutoFit, an expert system that automates the construction of the fitting model; AACanvas, for producing publication figures; and AAMultiLayer, which turns peak areas into the thicknesses and composition of a layered sample.

The emphasis is on XPS, where the method is most fully developed, but the peak-fitting approach and the tools apply equally to Raman and luminescence spectra, and examples of both will be shown.

 

Topics

 

• AAnalyzer — the essentials. Loading spectra, energy calibration, proposing components, choosing line shapes, and fixing, freeing, and correlating parameters. The Active Background Approach and why it matters for the areas you report. Simultaneous fitting of several spectra, and uncertainties on the fitted parameters.
• AAutoFit — the expert system. An inference engine that analyzes the spectrum, proposes an initial model, and then iteratively adds, removes, and constrains components through a set of physically grounded heuristic rules, driving a Levenberg-Marquardt optimizer. Objective, reproducible fits — with the analyst kept in the loop throughout.
• AACanvas — the figures. Building the figures for the paper directly from the fits: multiple panels, imported curves, error bars, and annotations.
• AAMultiLayer — from areas to a layer model. Fitting layer thicknesses and stoichiometric coefficients from peak areas measured as a function of take-off angle. Calculation of effective attenuation lengths, laterally coexisting layers, and how an AAnalyzer project feeds the depth model.
• Complete end-to-end examples on XPS, Raman, and luminescence spectra.

Bring your own data — we will analyze it live

 

This is what makes attending worthwhile. Participants are encouraged to bring their own spectra — XPS, Raman, or luminescence, whether routine or problematic — to work on during the workshop. Time will be set aside for real cases from the audience: a spectrum that will not close, a fit you are not confident in, an angle-resolved series you would like to model. You learn far more solving your own problem than watching someone else’s get solved, and one attendee’s case very often turns out to be everyone’s.

Simple two-column text files work well, as do VAMAS files and the text exports from commercial instruments. If you are unsure whether your format will load, bring it anyway.

 

Important: install the software BEFORE the workshop

 

The workshop is hands-on and meant to be followed on your own machine. Installing on the day costs everyone’s time, so please do it in advance.

• Download the software from https://xpsoasis.org/download
• AAnalyzer is essential — without it none of the demonstrations can be followed. Install it, open it once, and confirm that it launches.
• AAMultiLayer is needed for the last part of the workshop. It is easiest to install both in one sitting.
• Bring a Windows laptop and its charger. If your institution requires IT approval or administrator rights to install software, please start that process well ahead of time — at a national laboratory this is usually the step that takes longest.

Who should attend

 

Graduate students, postdocs, staff scientists, and technical staff who measure or interpret spectra, and users who bring samples to a spectrometer and have to defend the numbers that come out. No prior experience with AAnalyzer is required. What helps is having one of your own spectra on hand and a specific question about it.