In our lab we use a Thermo Scientific Nicolet iS50 FT-IR spectrometer equipped with PIKE VeeMax III for ATR experiments. We use python scripts for automatic data collection and spectra processing.

Requirements

• Python 3.8+
• Packages:
  • pandas
  • matplotlib
  • pybaselines

Install:

pip install pandas matplotlib pybaselines

---

Series spectra collection

spa_series.py is for collecting a series of spectra with fixed intervals.

Usage

In the folder you want to store the raw data, run:

python ./spa_series.py

It will ask for the interval in seconds and how many times you want to collect. It will generate *.spa files which are the raw data file from collections, and corresponding *.csv files for future data processing.

How it works

The script will first generate a macro file at every collection, which specify a series actions: collect one spectrum, save it as a processing.spa file, and save it as a processing.csv file. The script will then start the collection by running the macro, and after it detects the processsing.spa file, it will rename the processsing.spa as "order.spa" and rename "processsing.csv" as order.csv, and enter the next cycle of collection.

---

Data processing

All data processing functions are contained in the spectra_processing.py

What it does

• Combine a folder of CSVs (assumed format: wavenumber, intensity) into:
  • combined_raw.csv — all raw intensities side-by-side
  • referenced_raw.csv — each intensity minus the first file’s intensity
• Baseline-fit & subtract using methods from pybaselines
• Select a wavenumber window and subset of columns
• Plot selected columns with cosmetic options and save a PNG

Folder & data assumptions

• Place your input CSV files in a folder named raw/ in the current working directory.
• Each CSV must have two columns without headers:
  1. wavenumber (x), 2) intensity (y).
• The first CSV (alphabetical/OS order) is used as the reference spectrum for subtraction.

Tip: If your files have headers or different separators, adjust the pd.read_csv(..., header=None) line accordingly.

Quick start

Run the script directly to generate combined and referenced matrices:

python spectra_processing.py

This calls combining_series() with the default raw/ directory and writes:

• combined_raw.csv
• referenced_raw.csv

Functions (API)

combining_series()

Combines all .csv files in ./raw into two matrices:

• Inputs: none.
• Outputs (files):
  • combined_raw.csv with columns: Wave number, <file1>, <file2>, ...
  • referenced_raw.csv with columns: Wave number, <file1_minus_ref>, ...

bkg_fitting(fitter, x, y)

Fits a baseline to a single y series using pybaselines.

• Parameters:
  • fitter: one of "modpoly", "asls", "mor", "snip"
  • x: 1D array-like of wavenumbers
  • y: 1D array-like of intensities
• Returns: (bkg, params) where bkg is the fitted baseline

bkg_subtraction(df, fitter)

Fit the background using the fitter specified and baseline-subtracts all y-columns in a DataFrame.

• Parameters:
  • df: DataFrame where first column is x (wavenumber), remaining columns are spectra
  • fitter: one of the baseline methods above
• Returns: new DataFrame with the same columns where y → y - baseline

columns_selection(df, wave_range, cols)

Extracts a wavenumber window and a subset of columns.

• Parameters:
  • df: input DataFrame (first column must be wavenumber)
  • wave_range: tuple (xmin, xmax); if None, uses full range
  • cols: list of column names or indices
• Side effect: writes selected_(xmin, xmax).csv
• Returns: the filtered DataFrame

plot_columns(df, xlim=None, ylim=None, reverse_x=True)

Plots all y-columns vs. the first x-column and saves a PNG.

• Parameters:
  • df: DataFrame (first column is x)
  • xlim: (xmin, xmax) or None for full
  • ylim: (ymin, ymax) or None
  • reverse_x: True to invert the x-axis (common for wavenumber)
• Outputs:
  • Displays the plot
  • Saves spectrum_(xlim).png

Example workflow (interactive)

import pandas as pd

# 1) Combine and reference-correct
combining_series()
combined = pd.read_csv('combined_raw.csv')
referenced = pd.read_csv('referenced_raw.csv')

# 2) Select a region and subset of columns (e.g., columns 1..5)
sel = columns_selection(referenced, wave_range=(800, 1800), cols=[1,2,3,4,5])

# 3) Baseline subtract the selected region
sel_bkg = bkg_substraction(sel, fitter="asls")

# 4) Plot (Raman-style: decreasing wavenumber to the right)
plot_columns(sel_bkg, xlim=(800, 1800), reverse_x=True)

---

Common pitfalls & tips

• Column labels in plots:
  Legend label logic expects integer column names. Adjust if using filenames.
• Wavelength window save name:
  columns_selection writes selected_(xmin, xmax).csv; edit code for cleaner naming.
• Baseline parameters:
  The defaults (lam, p, poly_order, etc.) are generic. Tweak for your data.

---

Acknowledgments

Baseline fitting uses the excellent pybaselines library.