Structural Color from Cellulose Nanocrystals or Chitin Nanocrystals: Self-Assembly, Optics, and Applications

Widespread concerns over the impact of human activity on the environment have resulted in a desire to replace artificial functional materials with naturally derived alternatives. As such, polysaccharides are drawing increasing attention due to offering a renewable, biodegradable, and biocompatible feedstock for functional nanomaterials. In particular, nanocrystals of cellulose and chitin have emerged as versatile and sustainable building blocks for diverse applications, ranging from mechanical reinforcement to structural coloration. Much of this interest arises from the tendency of these colloidally stable nanoparticles to self-organize in water into a lyotropic cholesteric liquid crystal, which can be readily manipulated in terms of its periodicity, structure, and geometry. Importantly, this helicoidal ordering can be retained into the solid-state, offering an accessible route to complex nanostructured films, coatings, and particles. In this review, the process of forming iridescent, structurally colored films from suspensions of cellulose nanocrystals (CNCs) is summarized and the mechanisms underlying the chemical and physical phenomena at each stage in the process explored. Analogy is then drawn with chitin nanocrystals (ChNCs), allowing for key differences to be critically assessed and strategies toward structural coloration to be presented. Importantly, the progress toward translating this technology from academia to industry is summarized, with unresolved scientific and technical questions put forward as challenges to the community.


S1 DLVO theory for charged cylinders Pair interaction of charged rod-like CNCs
In DLVO theory as initially developed for spherical particles, the mutual orientation of the particles is irrelevant and the pair potential is determined by the distance  between the centers of mass of the particles, or the gap ℎ =  −  1 −  2 between the particle surfaces, assuming their radii are respectively  1 and  2 .For rod-like particles however, the interactions strongly depend on their mutual orientation.
Let us define the distance  axes as the shortest distance between the axes of two spherocylinders of radii respectively  1 and  2 , and lengths  ≫ ( 1 ,  2 ) oriented either parallel or perpendicular to one another, and the gap ℎ =  axes −  1 −  2 .
The pair interaction potential between rod-like particles in these specific directions can then be written as: where the two potentials correspond to parallel and perpendicular configurations, respectively.The attractive (van der Waals) and repulsive (electrostatic) contributions are evaluated in the following sections.

Attractive potential
The attractive van der Waals contribution  vdW () between two spherocylindrical particles of radii  1 and  2 , spaced from one another by a surface-to-surface distance ℎ, can be approximated by: 1 where  is the Hamaker constant, which is specific to the case of the materials of the particles 1 and 2 and the bulk medium in between.For CNCs in water, estimations of the Hamaker constant were estimated to  = 8 × 10 −21 J (from a film of cellulose II allomorph). 2 While this expression assumes spherocylindrical rods, the radii values relevant for CNCs can be approximated as  1 =  2 ≈ √/2, where  is the width of a CNC (typically 10-20 nm for cotton) and  its thickness (typically 5-7 nm for cotton).Note that there are alternative models for interactions between rods, such as that derived by Sparnaay, 3 and later corrected by Buining et al. 4 These models assume that the interior of the particle is made of a homogeneous, isotropic material.For refined geometrical models and anisotropic materials potentially more suitable for CNCs, the interested reader can find refined expressions of van der Waals potentials in this comprehensive handbook. 5

Repulsive potential
The repulsive electrostatic contribution  elec () between similarly shaped particles is given, under the assumption of large , in a 1:1 electrolyte (i.e., made of monovalent ions), by: 4 Here,  = 10 3    is the number density of ions (in #/m 3 ), where   is the Avogadro constant and  the ionic strength (in mol/L),   the Boltzmann constant,  the temperature,  the elementary charge and  0 is the surface potential.The Debye length  −1 is defined in section 5.1 of the main text.

S2 Analysis of Experimental Data from Angle-Resolved Optical Spectroscopy Specular Scan
The specular scan is sensitive to the domains pointing normal to the sample surface, and the wavelength of its highest reflection will follow Bragg's law for the pitch of the domains with no tilt: Where ′(0) is the pitch of local domains with tilt ′ = 0. Here,  is an integer corresponding to the diffraction order, with  = 1 for the main diffraction peak.As explained in section 7.3.5, higher orders are expected for non-normal incidence ( ext ≠ 0), and even at normal incidence if the domains are distorted (see section 7.3.6).

Scattering Scan
The reflected wavelength measured in a scattering scan,   (  ext ), is given using a parametric expression of   (′) and   ext (′) derived from the Fergason's law as: ext = arcsin[ ℎ sin( loc + ′)] (10) where the pitch variation ′(′) has to be accounted for using the expressions derived in section 6.1.

Tilt Scan
Finally, the reflected wavelength measured in a tilt scan,   (  ), is given as: where, again, the pitch variation ′(′) has to be accounted for.

Dispersion of Film Flakes in a Cylindrical Vial
For a suspension of CNC flakes freely floating inside a cylindrical vial, the tilt angle becomes irrelevant by symmetry, and the sample characterization is made by scanning the detector angle.
The normal incidence through the air-vial interface justifies discarding the Snell's law correction there, but the refractive index mismatch between the fluid and the flakes can still require Snell's law adjustment.The optical response is then well described by an adapted Fergason's law, where the local specular response of the films is dominating their individual off-specular contributions: loc =  ′ = ∆ ext (16 Where  is the pitch in suspension (expected to be a constant).

Cholesteric Suspensions in a Cylindrical Vial
For a cholesteric CNC suspension in a cylindrical vial, the sample characterization is made by scanning the detector angle, and the normal incidence through the air-vial and the vial-suspension interfaces allows for discarding Snell's law correction.The optical response is then given by Bragg's law: Where  is the pitch in suspension (expected to be a constant).

S3 Experimental Methods for Optical Characterization
This section provides the experimental and instrumental details used for the optical characterization of the example CNC film provided throughout the section 7.

S3.1 Angle-Resolved Optical Spectroscopy
Angle-resolved optical spectroscopy measurements were performed using a custom goniometer setup, which allowed free rotation of the sample and detector relative to the fixed illumination direction. 6A broadband xenon lamp (HPX2000, Ocean Optics) was coupled to a reflective collimator (RC08SMA-F01, Thorlabs) via a 200 µm fiber-optic cable (FC-UV200-2, Avantes) and used to illuminate the sample with a circular spot of diameter approx.2mm.Detection was performed using a second reflective collimator on the rotating detection arm, which was coupled to a UV-vis spectrometer (AvaSpec-HS2048, Avantes) via a 600 μm fiber-optic cable (FC-UV600-2, Avantes).The recorded light intensity was normalized to the specular reflection from a white Lambertian diffuser (WS-2, Avantes) measured at an angle of incidence of 5°.To accurately capture the scattered light across a wide range of intensities, the spectrometer integration time was adjusted automatically using a high dynamic range (HDR) method with integration times from 1.05 ms to 2000 ms.

S3.2 UV-vis Transmission Spectroscopy
Measurements were performed using the goniometer setup described above, with the detector facing the illumination direction.In this case the sample was illuminated using a 1000 μm diameter fiber-optic cable (FC-UV-1000-2-SR, Avantes) to provide a larger spot size (approx.15 mm).As the spot was larger than the aperture of the detection collimator, the directly transmitted light was focused using a plano-convex lens (LA1027-A, Thorlabs).To prevent saturation of the spectrometer, the transmitted intensity was reduced using a neutral density filter (ND20A, Thorlabs, optical density 2.0).Spectra are referenced to light transmission with no sample in the beam path.

S3.3 Double-ended Reflection Probe Spectroscopy
Measurements were performed using a double-ended reflection probe (R200-7-SR, Ocean Optics) as illustrated in Figure 40 of the main text.A broadband halogen lamp (SLS201L/M, Thorlabs) was used to illuminate the sample via six fiber cores arranged hexagonally, with the reflected light collected by a seventh central fiber and relayed to the spectrometer.Spectra were collected with a distance of 15 mm between the sample and probe, and reference to reflection from a silver mirror (PF10-03-P01, Thorlabs) at 15 mm.

S3.4 Optical Spectroscopy using an Integrating Sphere
For total reflectance and transmittance measurements, the sample was mounted onto an integrating sphere (LabSphere) as illustrated in Figure 40 of the main text.The sample was illuminated with a broadband xenon lamp coupled to a reflective collimator using a 200 μm fiberoptic cable, with a spot diameter of approx.5 mm.Diffuse scattered light from the sample was collected using a 1000 μm fiber-optic cable connected to a spectrometer.Total reflectance spectra were referenced to a Lambertian white diffuser, while total transmittance spectra were referenced to illumination without a sample.

S3.5 Polarized Optical Microscopy
POM was performed using a Zeiss Axioscope upright optical microscope in reflection and transmission modes, as illustrated in Figure 46 of the main text.Three objective lenses were employed: Zeiss EC Epiplan Neofluar 5x/0.13,EC Epiplan Apochromat 10x/0.30and EC Epiplan Apochromat 20x/0.60.Images were acquired using a digital CMOS camera (UI-3580LE, IDS), with the white balance set using a white Lambertian diffuser (WS-2, Avantes).Image magnification was verified using a microscope slide scale bar.
For unpolarized (UP) imaging, no polarizer or analyzer were used.For imaging between parallel polarizers (PP) and crossed polarizers (XP), the polarizer and analyzer were both broadband wire-grid linear polarizers (Thorlabs, WP25M-UB) arranged with polarizing axes parallel or perpendicular respectively.LCP and RCP imaging were performed with no polarizer in the light path, and an analyzer consisting of a superachromatic quarter-wave plate (B-Halle) followed by a linear polarizer arranged with the wave plate fast axis at 45° or 135° to the polarizer respectively.
Optical micro-spectroscopy was performed using the microscope setup in Figure 46 of the main text.A beamsplitter was used to split collected light between the microscope camera and a 100 μm diameter fiber-optic cable mounted in confocal configuration, allowing spectra to be obtained from a given spatial position and for a given polarization state.Reflection spectra were normalized to a silver mirror (Thorlabs, PF10-03-P01).UP spectra were normalized to a mirror illuminated in UP configuration.PP and XP spectra normalized to a mirror illuminated in PP configuration.LCP and RCP spectra were normalized to a mirror illuminated in the respective configurations, with separate references for LCP and RCP.