Four self-published papers by Ierotheos Katsoulis, from measurements made with a prism-coupling refractometer at the Electronic Devices and Materials Laboratory, University of West Attica. They cover polarization-resolved refractive indices of water, a polarization bias of the standard derivative method, instrument-corrected indices of turbid media, and what the same data can and cannot say about scattering.
The standard derivative method makes isotropic water look birefringent by +3 to +8 × 10⁻⁵. The size of the artefact follows from the measured peak shapes alone; corrected, |nTM − nTE| < 2 × 10⁻⁵.
Intralipid-20 % at 25 °C: n = 1.3691, 1.3595 and 1.3515 at 532, 964 and 1551 nm (± 0.002, ± 0.002, ± 0.0008), with TE and TM in agreement. Milk cream does not yield a reliable index by this method, and the paper shows why.
Fitting the whole reflectance edge shows that undiluted Intralipid scatters about half as strongly as scaled-up dilute data predict. It also shows the limits: n is good to about 10⁻³, k only to tens of percent, and the derivative method's k is biased low by back-scattered light.
The complete water analysis: TE/TM anisotropy, dispersion against the IAPWS formulation, and the temperature dependence of the index.
Upper bound on |nTM − nTE| for water at a prism interface, 532–1551 nm (95 % confidence).
Apparent birefringence the derivative method reports for isotropic water. It scales with (np/n)² − 1.
Index offset set by the direction of table rotation. Averaging both directions removes it.
How much the extinction coefficient would be overestimated if the artefact were read as absorption (water, 1551 nm).
Refractive index of Intralipid-20 % at 1551 nm and 25 °C, relative to air (± 0.0008).
Scattering of undiluted Intralipid-20 % relative to dilute measurements scaled up, at 964 and 1551 nm: the droplets are packed closely enough to scatter dependently.
How far the derivative method underestimates k in a turbid sample, because back-scattered light raises the edge. Its n stays within 5 × 10⁻⁴.
The download contains the four papers, all figures, the complete analysis code (Python), summary result tables, and a representative sample of 16 raw scans. A short script runs the analysis on the sample and prints the results next to the values in the papers; single scans agree to about 1 × 10⁻⁴.
The full measurement dataset was recorded at the Electronic Devices and Materials Laboratory (EDML), University of West Attica. It may be the property of the laboratory and is not disclosed. The sample files are shared for verification only and are not covered by the open licenses below.
→ Download package (zip, 5.1 MB)Measurements by Ierotheos Katsoulis at the Electronic Devices and Materials Laboratory (EDML), Department of Electrical and Electronic Engineering, University of West Attica, during an MSc by Research supervised by Prof. Konstantinos Moutzouris (2020–2021). Analysis and writing: Ierotheos Katsoulis, 2026. These are self-published preprints and have not been peer-reviewed.
Papers and figures are licensed CC BY 4.0: use them for anything, with credit. Code is MIT-licensed. To cite:
The other three papers are cited the same way, by title.