**2. Materials and Methods**

#### *2.1. Materials for Paper Coating and Coating Application*

An aqueous dispersion of hybrid organic nanoparticles was obtained by the imidization reaction of a given high-molecular-weight poly(styrene-*co*-maleic anhydride) or SMA copolymer (molecular weight *M*<sup>w</sup> = 80.000 mol, 26 mol % maleic anhydride), in the presence of ammonium hydroxide (NH4OH) and different types of vegetable oil. The resulting hybrid organic nanoparticles of poly(styrene-*co*-maleimide) or SMI with oil (SMI/oil) were obtained in combination with soy oil (SO), corn oil (CO), rapeseed oil (RO), sunflower oil (SfO), castor oil (CaO), and hydrogenated castor oil (HCO). More details about the reaction conditions and a full characterization of the aqueous nanoparticle dispersions as well as the dried nanoparticles can be found elsewhere [21]. Some characteristics of the oil qualities and SMI/oil nanoparticle dispersions are summarized in Table 1: the solid content (S.C.) was determined by infrared drying and weighing (LP16, Mettler Toledo, Greifensee, Switzerland); the viscosity was measured with a portable viscosity meter (Brookfield, DV-II Pro, Brookfield Engineering Laboratories, Middleboro, MA, USA), using a spindle n◦5 at rotation speed of 100 rpm, and the particle sizes were measured by dynamic light scattering (Nanosizer ZS90, Malvern, Malvern, UK).

**Table 1.** Characteristics of the oils and SMI/oil nanoparticle dispersions for paper coatings.


The SMI/oil nanoparticle coatings were deposited onto a paper substrate with a laboratory-scale K303 Multi-coater (RK Print Coat Instruments Ltd., Royston, UK). Two different metering bars were used at a constant speed of 6 mm/s, resulting in a thin coating (dry weight 4.0 ± 0.2 g/m2) and a thick coating (dry weight 6.0 ± 0.2 g/m2). The base paper sheets (100 g/m2, thickness 125 <sup>μ</sup>m, Mondi Business Paper, Vienna, Austria) contained bleached long-fiber and short-fiber kraft pulp with internal sizing and calendering. All coated paper samples were immediately dried in a circulating hot-air oven for 2 min at 120 ◦C, and stored in a controlled environment (23 ◦C, 60% RH) until further use. The characterization was done on the as-deposited coatings without any further processing (no calendering) to intentionally keep the specific micro- to nanoscale coating morphology.

#### *2.2. Paper Coating Characterization*

The scanning electron microscopy (SEM) was performed on a Hitachi Tabletop TM3000 microscope (Manufacturer, Krefeld, Germany) at different magnifications (800×, 2000×) and 15 kV voltage. An ultra-high-resolution image was obtained by FEG-SEM analysis using a FEI Nova 600 NanoLab focused ion beam workstation (Manufacturer, Hillsboro, OR, USA).

The confocal micro-Raman microscopy was performed on a dispersive Perkin Elmer Raman Flex 400 equipment (Manufacturer, Rodgau, Germany) with a multichannel charge-coupled array detector (CCD). The measuring conditions were optimized to avoid fluorescence and provide a good signal-to-noise ratio, using a of diode NIR laser (785 nm) with a maximum power output of 100 mW at the head and selecting a 40 mW measured laser power output at the sample position. The laser light was coupled to an optical microscope (Olympus BX51, Hamburg, Germany) equipped with a motorized piezoelectric *x*, *y* micro-Raman stage. The chemical Raman maps were recorded over a

surface area of 5 × 5 mm<sup>2</sup> using an objective lens of 20× (numerical aperture *NA* = 0.40) and a pinhole size of 50 μm. With a refractive index *n* = 1.5 for the samples, a lateral resolution of 2 μm and depth resolution of 5 μm was obtained. The Raman spectra were recorded at 100 × 100 points with 0.05 mm interdistance, 5 s exposure time, and six exposures per point (i.e., a total measuring time of about 14 h per map). The spectra were recorded at 3200–200 cm−<sup>1</sup> with 4 cm−<sup>1</sup> spectral resolution. The data were processed with Spectrum 10 analysis software (Perkin Elmer, Waltham, MA, USA) and Spectrum Image software (Version R1.7, Perkin Elmer, Waltham, MA, USA) to plot surface maps with average intensities, band ratio intensities, or single wavenumber intensities after normalization and baseline correction. Further processing of the Raman surface maps has been performed using the Image J processing software program (version 1.32j).

The statistical analysis of Raman spectra was done with Unscrambler 10.1 software. The input variables for the model were the baseline-corrected Raman spectra over the full wavenumber region. A calibration model for the principal component analysis (PCA) was developed by including 10 spectra per coating type (totally 60 spectra), recorded at random places over the coated surface area. The calibration model was verified by a cross-validation procedure, leaving out one of the calibration samples from the regression model and performing a new model for the remaining calibration samples. This validation method is considered when the number of samples is too small to have an independent training and validation set. The model was externally validated with an independent dataset (new coated paper samples), including five spectra per coating type that were recorded on independent surface areas.
