Nanostructure and anisotropy of 3D printed lyotropic liquid crystals studied by scattering and birefringence imaging

Extrusion-based 3D printing of hexagonal and lamellar lyotropic liquid crystals is a powerful technique to produce hierarchical materials with well-defined anisotropic structure. Tailoring the properties of 3D printed objects requires a precise control of the nanostructure; however, a sufficiently high degree of anisotropy is often not achieved. In this study, scanning small angle X-ray scattering was performed in situ at the exit of the needle during 3D printing. We study the induced anisotropy and nanostructure in hexagonal and lamellar lyotropic liquid crystals. Mapping of extruded filaments during printing revealed that narrower nozzle diameters (370 µ m) resulted in less anisotropic structures with a wider distribution of orientation angles across the cross section, while larger nozzle diameters (550 µ m) resulted in more anisotropic structures with an overall higher degree of orientation. The apparent wall shear rate is higher for the narrower nozzle, which produces wall slip, resulting in a highly anisotropic shell, and a less aligned filament core. Further examination of the filaments revealed phase transitions due to solvent evaporation. The time scales were of 10 – 20 min of exposure to atmospheric conditions. Simultaneously, a loss in the macroscopic anisotropy of the hexagonal self-assembled structure was observed. These processes occur during and after extrusion-based 3D printing of liquid crystals and limit the fine control of the final structure. The variability of structures achieved for our different systems highlights the importance of structural characterization during and after extrusion to guarantee high anisotropy and well-defined structures.


Introduction
In additive manufacturing, also known as 3D printing, objects are manufactured layer-by-layer based on a Computer-Aided Design (CAD) model, thereby enabling the fast production of complex geometries. In addition, it does not require the use of molds or dies and does not generate excess waste material, thereby attracting attention from both temperature [4], UV light [5], time [6], or sonication [7], as well as the opportunity to create complex functional systems. A wide variety of systems, including hydrogels, colloids and liquid crystals, have shown to be suitable materials for extrusion-based 3D printing intended for applications such as optically responsive devices [8], creation of composites with complex hierarchical structures [9], bio-inks capable to include living cells in their composition for tissue engineering [10] and designed for drug-delivery [11], among others.
For inks to be suitable for extrusion-based 3D printing, they should have well-controlled rheology, be designed for specific printing parameters and for a particular final application. The intrinsic properties of the material and its response to the imposed forces by the 3D printing setup and parameters will define the final structure and properties of the ink. It has been shown that inks having a non-Newtonian behavior with shear thinning and yield stress, such as silicon inks [12] and graphene oxide suspensions [13], are generally desired in 3D printing. The yield stress must be high enough so that the extruded filament and the final 3D object keeps its shape before crosslinking processes can be performed. However, high yield stresses might lead to clogging, or require high pressures. Therefore, low yield stresses with shear thinning behavior facilitate the flow through the nozzle. Furthermore, conditions for particle orientation during 3D printing have been assessed for cellulose nanocrystal suspensions, which rely on inducing stresses above the yield stress of the sample [14]. Particle orientation dynamics is determined based on the applied deformation rate and specific particle network interactions.
Among different inks, amphiphilic block co-polymers can assemble in ordered structures when they are in contact with a polar solvent. The hydrophobicity of certain blocks creates a segregation process, in which several blocks associate together to form an apolar domain while the hydrophilic block(s) create a polar domain. This leads to the creation of self-assembled structures such as spheres, bilayers, or cylinders [15], as schematically shown in Fig. 1a for cylinders and bilayers. Under certain conditions of temperature and polymer concentration in one or more solvents, these block co-polymers form lyotropic liquid crystals, which are made of repetitive self-assembled units with order in the long range in a well-defined crystalline lattices [16]. These lattices are highly dependent on the composition, which gives possibilities to study inks of different self-assembled structures for use in drug delivery [17] and as biomimetic composites [18], for example. A precise control of the anisotropy is an important requirement for the successful performance of hierarchical materials [19,20], optimizing the 3D printing process towards a high degree of orientation and uniformity across the strand is therefore needed. Controlling the anisotropy at the nanoscale enables the creation of hierarchical materials with well-defined anisotropic physical properties. These materials are of critical importance in heart valve replacements, where a fibrinous flexible tissue with anisotropic mechanical properties is needed [21], or to direct the mineralization process leading to an anisotropic mineralized composite [22].
Scattering techniques are appropriate to study the nanostructure and anisotropy of the intrinsic order of liquid crystals [23]. The use of small angle X-ray scattering (SAXS) is suitable since the building blocks of the Fig. 1. a) Scheme of the self-assembled hexagonal and lamellar lyotropic liquid crystals and experimental set-up of the scanning SAXS experiment using in situ 3D printing. A 40 µm X-ray beam scans the extruded polymer and the scattering signal is recorded by a detector. Radial integration of the scattering signal for the polymer F in hexagonal self-assembled structure (b), P in hexagonal (c) and P in lamellar (d) printed with a 460 µm nozzle. The highlighted regions represent the selected q-range used in the calculations. A representative scattering pattern is shown as an example for each liquid crystal. q* references the value of the main diffraction peak. The distance between cylinders (d) and lamella (D), respectively, is calculated from the main diffraction peak. liquid crystals are in the range of tens of nanometers, and the created order produces a strong diffraction signal characteristic of each self-assembled structure. A focused beam enables the study of hierarchical 3D printed materials, allowing for spatial resolution in the macroscale by scanning the printed filament through the beam. In addition to scattering, the orientation in liquid crystals can be probed by monitoring changes in the polarization angle of light. The birefringence of polymers and macromolecules have two contributions; intrinsic and form birefringence [24]. The intrinsic birefringence is caused by the optical properties of molecules, whereas the form birefringence is caused by an anisotropic aggregate shape, such as cylinders or worm-like micelles [25]. In liquid crystals, isotropic phases, such as the cubic micellar phase, do not show birefringence, whereas hexagonal or lamellar phases typically show birefringence. However, not all anisotropic phases show a measurable birefringence, as the intrinsic and form birefringence can have opposing signs [26,27].
Control of macroscopic anisotropy is often desired since it affects electrical and thermal conductivity, optical properties, mechanical performance, or magnetic properties [28]. Anisotropy can be introduced by the shear and extensional forces during 3D printing which can reach relatively high alignment [29]. However, alignment using shear is limited to the printing direction and the shear forces created are limited by the nozzle diameter and flow rate in it [30]. This method, apart from reaching high degree of anisotropy, can be used to generate a specific texture in the nanostructure which could be used, for instance as templating to align particles [31]. However, knowledge to maximize the anisotropy and understand the source of the defects during and directly after printing is yet to be achieved. By combining SAXS and birefringence microscopy, we hereby aimed to visualize and explore the processes involved in the alignment of Pluronic based lyotropic liquid crystals in normal hexagonal and lamellar phase.

Scanning SAXS
Scanning SAXS was performed at the coherent small-angle X-ray scattering (cSAXS) beamline at the Swiss Light Source (SLS) Paul Scherrer Institute (PSI, Switzerland). A focused X-ray beam of 40 × 40 µm was used with a photon energy of 11.2 keV (1.107 Å), selected by a fixed-exit double crystal Si(111) monochromator. The 3D printing cartridge and nozzle were mounted on a motorized stage with two degrees of freedom in the x-y plane, perpendicular to the beam, allowing to scan the 3D printed filament through the beam. A flight tube with a length of 2 m was placed between the sample and the detector to minimize air scattering and absorption. The scattering signal was recorded by a Pilatus 2 M detector [32] and the transmitted beam was measured with a photodiode on a beamstop inside the flight tube. The exposure time for all the in situ measurements was 0.08 s, at a sample-to-detector distance of 2.158 m, which was calibrated with silver behenate.
The scattering signal was azimuthally integrated and the phase of the liquid crystals identified by the ratio between the main diffraction peaks, corresponding to the distance between cylinders or bilayers in the crystals [15]. The angle of orientation and orientation parameter were calculated using the Hermans' Algorithm [33], which defines the Hermans' orientation parameter 〈 f a 〉 as in Eq. (1), where I(ϕ) is the intensity for each azimuthal angle (ϕ) [34]. The Hermans' orientation parameter is a broadly used scalar order parameter, which defines the orientation of particle assemblies from X-ray scattering experiments using the intensity of the scattering over the azimuthal angle. The azimuthal intensity was normalized by its maximum intensity and interpolated using a Lorentzian curve. The angle of orientation was obtained by the peak position in the azimuthal intensity after the interpolation indicating the angle in real space, color-coded according to the displayed color-wheel.
The orientation parameter and angle of orientation in the scanning maps were averaged in the lowest 500 µm in order to avoid artifacts in the vicinity of the nozzle, such as instabilities and material expansion or contraction.

Birefringence microscopy
Spatially resolved retardance and angle of the optical fast axis were measured with the Exicor Birefringence MicroImager™ (Hinds Instruments, Inc., OR, USA). All measurements were performed with a LED source having a wavelength of 475 nm. The optical set-up consisted of a linear polarizer at 0 • , a photoelastic modulator (PEM) at 45 • , a PEM at 0 • and a linear polarizer at 45 • . The samples were placed in the sample stage between the two PEM and monitored with an objective of 20x (FH 1 and PH 1 ), which creates a field of view of 500 × 500 µm with a spatial resolution of about 0.25 µm/pixel. For the PL α , the samples were imaged with a 10x objective which has a field of view of 1 × 1 mm and a spatial resolution of 0.50 µm/pixel. The microscope has a 2048 × 2048 pixel CCD camera. A stroboscopic light source was used to calculate the Mueller matrix components and provide quantitative measurements of the sample retardance and angle of the optical fast axis in the field of view [35,36]. The angle of the fast axis is color coded for easier interpretation.

Table 1
Description of the prepared samples including composition, self-assembled structure, and characteristic distance between cylinders (d) and lamellae (D).

Self-assembled nanostructure of the 3D printed filaments
Two different triblock co-polymers with the same block components, poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) (PEO-PPO-PEO) but with different hydrophilic chain lengths were used in this study; polymer F with a block composition of PEO 100 PPO 70 PEO 100 and polymer P of PEO 20 PPO 70 PEO 20 . Three different lyotropic liquid crystals were formed by adjusting the ratio between the polymer and the solvent (s) (see Materials and methods, Table 1). The resulting gels were extruded through a 3D printer cartridge mounted on motorized stages to scan the obtained filament through the X-ray beam, as schematically shown in Fig. 1a. Representative radial integration of the scattering signal for all the samples are shown in Fig. 1b-d. The obtained diffraction peaks proved that the materials had a mesoscopic order with two different self-assembled structures. For the liquid crystals made of polymer F with composition polymer:water:butanol 35:50:15 and P with 45:55:0, the diffraction peaks appear at the defined positions in the scattering vector q with a ratio of 1:√3:2:√7:3 (Fig. 1b, c), which is characteristic for a normal hexagonal compact system (H 1 ). Those two samples were therefore labeled as FH 1 and PH 1 , respectively. In a hexagonal phase, the polymer assembles in a cylindrical arrangement with the hydrophilic polyethylene chains facing the outer aqueous continues media and the middle hydrophobic polypropylene block facing the internal apolar domains. The cylinders are self-assembled in a hexagonal lattice with cylinder inter-distance of 167.1 Å and 151.4 Å for FH 1 and PH 1 , respectively, as determined from the first diffraction peak. In the hexagonal phase, the aligned equally-spaced cylinders form an anisotropic scattering pattern, perpendicular to the long axis of the cylinders, i.e. in the direction of the alignment. The polymer P with composition 75:25:0 had the diffraction peaks at q-values with a ratio of 1:2:3 ( Fig. 1d) characteristic for a lamellar structure (L α ) where the polymer form bilayers in a lamellar stack. This sample was denoted as PL α , and had a lamellar distance of 113.9 Å. The lamellar phase also presents anisotropy in the scattering signal, where lamellae align parallel to the walls and in the direction of extrusion. The symmetry plane of the arcs is parallel to the lamella plane. Fig. 2 summarizes the scanning SAXS results of the in situ 3D printing experiments, showing the Hermans' orientation parameter [33] 〈 f a 〉 on the left side and the angle of orientation in the right side of the map for three different inner nozzle diameters; 370 µm, 460 µm and 550 µm as indicated with the red line on top, which are expected to produce a cylindrical filament. The high viscosity of PH 1 did not allow for extrusion through the smallest nozzle diameter and therefore it is not displayed in the figure. Both hexagonal structures had a higher orientation parameter when extruded through a larger nozzle diameter, as shown in the grey scale representation of the scanning maps (Fig. 2). The mean orientation parameter had an increase of 30% for FH 1 and 38% for PH 1 from the smallest to the largest nozzle, which reach a value up to 〈 f a 〉 FH1 = 0.64 ± 0.06 and 〈 f a 〉 PH1 = 0.61 ± 0.02, respectively. The overall orientation parameter in the lamellar structure, PL α , is in contrast not significantly influenced by the nozzle diameter, as observed in both the scanning map in Fig. 2f-h as well as in Fig. 3, where each scanning line of the filament was radially averaged from the center in the lower 500 µm of the map. Most filaments had a core-shell structure where the outer shell had a higher orientation parameter than the core of the filament. FH 1 had a core-shell structure for all three nozzles sizes, but the effect is less pronounced for the largest diameter (yellow stars in Fig. 3a) where the orientation parameter was found to be higher throughout the whole filament compared to smaller nozzle sizes. PH 1 exhibited a thin highly aligned shell (Fig. S2), when extruded through the 460 μm nozzle while becoming highly homogeneous when using the 550 µm nozzle (Figs. 2de and 3a). The bright pixels in PH 1 are indeed showing a highly oriented hexagonal phase and is not due to an effect of the edge scattering (see Fig. S2). The orientation parameter in the lamellar structure, PL α has a more gradual change from the core to the shell, but is not significantly influenced by the nozzle diameter, as observed in Fig. 3a.

Macroscopic anisotropy of the 3D printed filaments
The angle of orientation induced by extrusion during the 3D printing process is color-coded according to the color wheel and indicated by black lines shown in Fig. 2. The mean angle of the main scattering orientation is in the range ±5 • in all cases, defining 0 • in the filament axis (vertical direction). This corresponds to an orientation of hexagonally packed cylinders with their long axis in the direction of extrusion (Fig. 2a-e), and a stack of lamellae in a coaxial wrapping following the filament axis (Fig. 2f-h). It is important to mention that the orientation of the anisotropy observed in the lamellar phase is always perpendicular to the vector normal to the surface of the lamellae (director), commonly used to define the orientation of the lamellar structures. The variation of the angle of orientation across the filament is strongly dependent on the nozzle diameter (Fig. 2). In Fig. 3b, 0 • corresponds to an orientation in the extrusion direction, the deviation from it plotted in absolute values to account for the radial symmetry. The filaments produced by the largest nozzle (550 µm) had a more homogeneous alignment in the extrusion direction (turquoise) for the three materials as also seen in Fig. 3b with an angle of orientation close to zero. Smaller nozzles created a distinct pattern of alignment where the nanostructures are directed diagonally downwards from each side of the wall (blue and green, respectively). The 460 µm nozzle created a rather abrupt change in the middle forming a V-shape, while for the smallest nozzle (370 µm), structures in perpendicular orientation to the extrusion direction (red/ pink) are observed, in particular for the lamellar phase where the transition is rather smooth forming a U-shape (Figs. 2f and 3b PL α -370 µm). The diameter of the filaments produced by the hexagonal structures is very similar to the inner diameter of the nozzle indicated with the red line in Fig. 2, and no change in diameter is observed along the printing direction. However, the lamellar phase PL α exhibits an expansion by 56 -73% after the exit of the nozzle, creating a filament bigger than the inner diameter of the nozzle (Fig. S3). This has an effect on the nanostructure orientation as seen in Fig. 2f-h, which are in the first 150-400 µm pointing outwards before they reverse into the U-or Vshapes described above. The expansion was not related with any change in the nanostructure since no change in lamellar spacing could be observed in the scattering signal (see Fig. S4).
The core-shell structure can be explained by a higher local shear rate in the proximities of the walls than in the middle, which significantly influences the viscosity and the orientation parameter in lyotropic liquid crystals [37]. In addition, an apparent wall slip effect could be detected in the liquid crystals in simple shear flow for shear rates above γ = 0.01 s -1 . Wall slip could occur towards the end of the nozzle as the isotropic pressure approaches atmospheric pressure conditions. Wall slip is observed in fluids with a large yield stress such as highly concentrated suspensions, caused by large shear rates, which produces a highly oriented outer layer in the filament [38,39]. The internal unyielded core is then pushed by the surrounding material. Narrower nozzles increase the wall shear stress likely increasing the wall slip effect [40][41][42][43]. This has revealed itself in a three-zone velocity profile for 3D printing inks based on colloidal gels [44]. In the core of that filament, an unyielded gel moves at a constant velocity (plug flow). The additional swelling effect observed in the lamellar phase, PL α , could be related with the absence or diminished presence of apparent slip. High wall shear stress creates a highly aligned layer that might relax producing a rearrangement and swelling of the filament. In summary, the viscoelasticity of the material plays a very important role in the homogeneity and influence of the printing nozzle diameter. The use of larger nozzle diameters achieved higher degree of anisotropy, i.e., higher orientation parameter and more homogeneous distribution of angles.

Time evolution of the anisotropy
The structure of inks containing volatile solvents is susceptible to evaporation during the manufacturing process. The evaporation may change their properties due to change in composition, which is in the case of lyotropic liquid crystals crucial to define their self-assembled structure. To shed light on phase stability in an open environment, birefringence microscopy was performed on the 3D printed filaments, which were imaged directly after printing and over a period of 30 min with one capture per min. The filament was extruded in the vertical position through a nozzle with inner diameter of 460 µm and transferred immediately to a glass slide for imaging. The change in the optical anisotropy of the ordered polymeric structures made it possible to follow the evolution of the anisotropy with time [45]. This measurement is equivalent to the waiting time the filament remains on the printing surface before the object is completed and the structure is fixated by crosslinking. Fig. 4 shows the angle of the optical fast axis of the 3D printed liquid crystals over time according to the displayed color wheel and the value, i.e. the darkness, scaled with the retardance (separated values of the angle and retardance are plotted in the Fig. S5). The optical fast axis of FH 1 in Fig. 4a has an orientation parallel to the printing direction (Fig. S5a), which is in agreement with the scanning SAXS experiments shown in Fig. 2a-c. A highly aligned filament is initially printed with low retardance (Fig. S5b) i.e., low optical anisotropy, which at first increases with time. After about 17 min, the macroscopic orientation is lost and the formation of microdomains is observed, starting with a few microdomains of different orientations before extending to the whole analyzed area. The formation of microdomains goes together with a decrease in the retardance after 20 min. The extrusion process through a nozzle increases the shear stress and fluid velocity considerably. The high stresses induced in the microstructure may eventually relax forming ordered microdomains with random orientations [46]. PH 1 in Fig. 4b, produced a filament with high anisotropy. A homogeneous angle of the fast axis and a high retardance is visible (Fig. S5c,d). The retardance decreases gradually in the first 4 min, presumably by the effect of evaporation and relaxation. Already in the first frame a distinct region appears in the outer layer of the filament, which grows in the first 5 min. A lower retardance and opposite angle than the core is observed. After 5 min, the retardance values reach a minimum and the intensity of the birefringence signal is too low to consider the measured angle of the optical fast axis as reliable. This new signal has a high similarity with the lamellar phase in PL α (Fig. 4c). The birefringence signal of the lamellar phase aligned radially in the cylinder shows a very low retardance with a homogeneous angle equal to the observed signal after the transition in PH 1 (Fig. S5e,f). The values of the retardance are too low to consider the angle accurate. The low retardance value of the lamellar phase is contradicting the high orientation parameter measured with SAXS. Macromolecules forming a bilayer have in general a positive intrinsic birefringence [24], whereas the form birefringence of lamellar systems is dominated by the arrangement and lamella interspace and is negative [47]. Certain configurations produce an interaction between the intrinsic and form birefringence that cancel each other, giving a net value close to zero [26]. The opposing effect of intrinsic and form birefringence can also explain the variation of the retardance within the two hexagonal phases, for which the intrinsic birefringence for molecules perpendicular to the long axis can be expected to be negative, whereas the form birefringence of a cylinder is positive [27]. The form birefringence is influenced by the d-spacing which may change with evaporation.

Time evolution of the self-assembled nanostructure
In order to further investigate the observed changes in the birefringence over time, the scattering signal of 3D printed filament was collected in situ, scanning across the diameter of the filament through the x-ray beam at several time points and then averaged for each scan line. It should be noted that the environmental conditions, such as humidity, at the beamline during the SAXS experiments were different than in the birefringence microscope, thus the exact times are not  (Fig. S5). The filament was 3D printed in the vertical axis and deposited on a glass slide in atmospheric conditions for the measurement. The time points, indicated in min, with most significant changes were selected to visualize the evolution. The background (glass slide) was masked. The scale bar represents 400 µm.
comparable, but the same sequence can be expected. For FH 1 in Fig. 5a the initial peak ratio is 1:√3:2:√7, representative of a hexagonal selfassembled structure. The hexagonal fingerprint remains visible over time, but the peaks get broader and less pronounced. In addition, there is a second population at lower q-values with a broader peak width at a characteristic distance of 167.41 Å. This can be related to the ordered domains in random directions observed by birefringence microscopy, perhaps in the interspace between ordered domains. In contrast to the retardance, the orientation parameter of the scattering pattern stays constant, with only a minor trend toward lower values observed with time, as shown in Fig. S6. Fig. 5b shows the time evolution of PH 1 starting with the characteristic hexagonal peak ratio 1:√3:2:√7. The main peak from the plane (01) is shifted to larger q-values and the secondary and tertiary peaks faint. A new structure appears with a peak ratio of 1:2:3 characteristic for a lamellar structure. Since the experiment was performed in atmospheric conditions, we assume a change of composition due to the loss of components by evaporation, which results in a shift in the phase diagram [15]. According to the phase diagram the loss of water displaces the composition towards the stability field of the lamellar phase. This phase transition was also observed by a change in the birefringence signal matching the lamellar phase as seen in Fig. 4c. A decrease in the orientation parameter is also observed in Fig. S7, in this case following the decrease of the retardance (Figs. 4b and S5d). The evolution of the sample PL α is shown in Fig. 5c. The initial structure is a lamellar stack with peak relationship 1:2:3. The effect of time and evaporation induced here a partial phase change with three prominent peaks appearing at low q-values. Together with the new peaks, a shoulder at low q appears which can be fitted with a form factor from a sphere in cubic arrangement with a radius of 136 Å (Fig. S8). Among the possible self-assembled arrangements of spherical micelles, a cubic structure with the space group Pm3n (8 micellar units per unit cell) is the most common structure observed in surfactants with low molecular weight [48]. As described by B. Svensson et al. [49], the assumption of two or even four micellar units per cubic cell, as in BCC and FCC, is incompatible with the estimations of the area of the polar-apolar interface. This assumption results in a lower area than the hexagonal self-assembled structures, which should be lower. The indexation of the diffraction peaks could presumably match the reflection of the crystalline planes (200), (210) and (211) as described in a similar system [49], however the reflection (110) is not present. This result suggests the coexistence of two phases, which evolve over time. The orientation parameter in the SAXS experiments does not significantly change, as seen in the scattering signal in Fig. S9. The phase diagrams represent the self-assembled phases for the given parameters in the thermodynamic equilibrium [15]. The performed experiments involving extrusion and quick environmental changes are far from the thermodynamic stability, which open the possibility for different phases to occur. The scattering signal was checked at different positions across the filament diameter (Figs. S5, S6 and S9). No significant changes were observed. The step size used in the scanning SAXS experiments does not allow to resolve small changes on the surface; however, it is visible using birefringence microscopy, as observed in Figs. 4 and S5. This finding brings attention to the fact that the structure must be checked after 3D printing and the fixation needs to occur at time scales smaller than any observed relaxation or phase change. The printing of large objects implies long times, in which the composition of the gels might change and therefore their structure. This could be counteracted by a controlled atmosphere during printing.

Conclusions
In this work, the anisotropy created by extrusion-based 3D printing was visualized in situ using scanning SAXS and birefringence microscopy. The angle of orientation and degree of anisotropy were measured in the aligned materials oriented in the flow direction. The general orientation of the structure was found to be in the printing direction. Surprisingly, a higher and more homogeneous orientation was achieved using larger nozzles. A core-shell structure was found in the orientation parameter with highly oriented outer layer and low oriented core. The use of smaller nozzles produced filaments with a wider distribution of angles, which make them less suitable for aligned filaments with controlled anisotropy.
The time evolution of the printed filament in ambient conditions was tracked by SAXS and birefringence imaging concerning the selfassembled nanostructure and alignment. A general trend towards relaxation and disorder was observed after long exposure to ambient pressure and atmospheric conditions. The effect of evaporation led to a phase change from hexagonal to lamellar in the self-assembled hexagonal liquid crystal (PH 1 ) and to a phase coexistence between lamellar and cubic in the lamellar PL α . The orientation visualized with scanning SAXS and birefringence microscopy was qualitatively comparable in the hexagonal phases, the lamellar phase however did not show birefringence. The Hermans' orientation parameter and the retardance had qualitatively similar trends, while in the retardance additional effects from the phase transition play an important role. The complexity of these self-assembled materials makes a detailed study of the behavior of each individual structure necessary in order to avoid undesired secondary effects, such as the observed phase transitions and changes in the orientation. These findings point to the importance of a detailed observation of the nanostructure in self-assembled systems such as lyotropic or thermotropic liquid crystals, in which an uncontrolled change in the environmental conditions may lead to a change in the desired material and alter its performance.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.