PEAXACT Tour Part 4 – Integrating Peaks
It is time to create a simple Integration Model to calculate peak areas, run the integration analysis, and introduce the Report Window.
Integration Model
Model building takes place in the main window and comprises three major steps:
- Defining a Pretreatment Model (optional, see previous part).
- Creating a Spectral Model that converts the spectrum into a quantity of interest.
- Defining a Calibration Model (optional). You will learn more about calibration in Part 6 of this tour.
The Integration Model is one type of spectral model that calculates areas by integrating the signal within specified ranges.
The area of a component peak is related to the component's concentration, which makes the Integration Model the simplest model for monitoring concentration changes. Creating the Integration Model in PEAXACT is also very simple: Select a spectrum and then use the Plot Panel to interactively add Integration Model Peaks, move integration boundaries, or change the baseline. The area is shown as a colored patch.
Integration Analysis
Do you have a lot of samples to analyze? Then you certainly don't want to build a model for each sample, but create one that works for all, including future samples yet to come. A good Integration Model only needs a few known spectra to create it, e.g., for finding the integration limits, and then can be used to analyze many unknown samples.
In case you have already measured some spectra, you can analyze them afterwards by running an offline Integration Analysis from the PEAXACT main window. Results are displayed in a Report Window.
The Report Window provides graphical and tabular reports that can be customized and exported to image and text files. Report customization makes use of sample features, e.g., to plot the results against time.
Another use case would be the deployment of the model to a field instrument and automatically trigger the analysis in real-time when a new spectrum is recorded. In this case, results are typically sent to different outputs like result files, OPC servers, or real-time charts. You will learn more about this in a later part.
Integration Models are great as long as component peaks don't overlap. In the next part, we start looking into spectral Hard Models, which can deal with overlapping peaks.