Time dependent lyotropic chromonic textures in PDMS-based microfluidic confinements

Nematic and columnar phases of lyotropic chromonic liquid crystals (LCLCs) have been long studied for their fundamental and applied prospects in material science and medical diagnostics. LCLC phases represent different self-assembled states of disc-shaped molecules, held together by noncovalent interactions that lead to highly sensitive concentration and temperature dependent properties. Yet, microscale insights into confined LCLCs, specifically in the context of confinement geometry and surface properties, are lacking. Here, we report the emergence of time dependent textures in static disodium chromoglycate (DSCG) solutions, confined in PDMS-based microfluidic devices. We use a combination of soft lithography, surface characterization and polarized optical imaging to generate and analyze the confinement-induced LCLC textures, and demonstrate that over time, herringbone and spherulite textures emerge due to spontaneous nematic (N) to columnar M-phase transition, propagating from the LCLC-PDMS interface into the LCLC bulk. By varying the confinement geometry, anchoring conditions and the initial DSCG concentration, we can systematically tune the temporal dynamics of the N to M-phase transition and textural behaviour of the confined LCLC. Since static molecular states register the initial conditions for LC flows, the time dependent boundary and bulk conditions reported here suggest that the local surface-mediated dynamics could be central in understanding LCLC flows, and in turn, the associated transport properties of this versatile material.


Introduction
Lyotropic chromonic liquid crystals (LCLC) are a class of lyotropic liquid crystals (LCs), that are formed by anisotropic assemblies of water-soluble disc-shaped molecules that have an aromatic core surrounded by ionic groups. Most commonly used chromonic LCs are sunset yellow, an azo dye used as food additive and disodium chromoglycate (DSCG), a anti-asthmatic drug. Unlike lyotropic LCs, LCLCs do not form micelles, rather they stacks up as linear aggregates, held together by non-covalent interactions that lead to self-assembled nematic (N-phase) or columnar (M-phase) with hexagonal arrangement [1,2,3,4]. In both N-phase and M-phase director is parallel to the columnar axis of the stacks, as was revealed by X-ray studies. The nematic phase consists of short columnar stacks of DSCG molecules and appears at room temperature for lower DSCG concentrations, whereas with increase in DSCG concentration the stacks assemble into a two dimensional hexagonal array of extended columns in the M-phase [5,6,7], archetypal herringbone and spherulite textures spanning a wide range of concentration and temperature conditions. The weak interaction forces underpinning the self-assembled phases render LCLCs highly responsive to external stimuli (temperature, concentration, pH, ionic content etc.) and geometric constraints [1,2,3,4], thus bestowing distinctive properties such as negative Despite continued efforts to understand LCLC rheological properties, motivated by both fundamental and applied interests, we still lack a mechanistic insight into the LCLC hydrodynamics at microscales -a universal variable in medical, industrial and biological settings. With the advancement in soft lithography techniques which has enabled precise design and modulation of microfluidic geometry, pressure and boundary conditions, both biological and material technology have seen major advancements over the last decades [35]. Recent progress in liquid crystal microfluidics has demonstrated how hydrodynamics, in combination with surface anchoring and confinement can be harnessed to generate tunable flow and topological structures with potential for novel applications. The work of Sengupta et al. [36,37,38] and others [39] have shown that by precisely tuning the flow of nematic LCs, anchoring conditions at channel walls and confinement conditions in microchannels interesting phenomenon such as defect-mediated flow at low Reynolds's number [40], non-Poiseuille flow profile [41], hydrodynamic cavitation [42] and crosstalk between topological field [43] are observed. The work on nematic LCs in microconfinement has shown significant insight on the interplay of topological defects with tunable microflows of nematic LC and finds applications in micro cargo systems [44,45], optofluidic system [46,47,48], and fluidic resistance circuits [49]. Based on these works, recently, flow-induced deformations have also been studied in chiral nematic LCs [50,51] and blue phases. Spontaneous tunable flow structures [52,53] were given a second look byČopar and co-workers [54], revealing a hidden non-equilibrium chiral nematic state that could be stabilized by tuning microchannel dimensions and the driving pressure.
The behaviour of LCLCs in strictly microfluidic environments still awaits exploration, with multiple key variables (that determine their interactions) yet to be uncovered and understood. As microfluidic environments offer distinct interplays between the surface, confinement and viscous parameters, one expects rich and unique interactions emerging at the LCLC-microfluidic interfaces. By substituting thermotropic nematic LCs with LCLCs within microfluidic confinements, the Authors have now opened a Pandora's box wherein a wealth of novel phenomena are being observed which warrant fundamental investigations. Relying on the Authors' insights from previous experiments on thermotropic LCs in strictly microfluidic confinements, here we continue with state-of-the-art soft lithography techniques to precisely tune the dimensions of the microchannels, and modify the surface anchoring conditions on the channel walls. It is known from study of thermotropic LCs in microfluidic environments that the initial conditions for flow-induced director response is defined by the static (no flow) conditions [53]. With a vision to comprehensively understand the flow-confinement-anchoring couplings in LCLCs confined within microchannels, here we analyze the fundamental no flow state of LCLCs wihtin microfluidic enviroments. Specifically, we characterize how LCLC self-organizes within microscale confinements as a function of three fundamental variables: (i) channel dimensions, (ii) anchoring conditions, and (iii) LCLC concentration. In doing so, we have uncovered a fourth factor -the experimental timescale -which plays a critical role establishing the static director configurations within PDMS-glass microchannels. Our results reveal that the set of the four variable are fundamental in determining the outcome of flow experiments (flow results will be discussed elsewhere), and thus need to be taken into account when designing LCLC microfluidic experiments. In our first attempt to bridge the gap between LCLC dynamics and microfluidics, here we will focus on recent results on the temporal dynamics of LCLC textures formed by DSCG solutions confined in PDMS-glass microchannels, spanning different aspect ratios (AR = width/height = w/h) and anchoring conditions. The basic structure of DSCG, along with the stack configurations representing different phases are shown in Figure 1.

Materials and Methods
Using a combination of precision microfluidics and polarization microscopy, we characterized the concentrationdependent static LCLC textures. We compared three concentrations of DSCG and characterized the LCLC optical textures as a function of anchoring conditions and channel aspect ratios, using polarizing optical microscopy and image analysis. Our results indicate that the emergence and orientation of the LCLC micro-domains are highly sensitive to confinement length scales and surface anchoring conditions. Additionally, a highlight of this work is the spontaneous nematic to columnar (M) phase transition in PDMS-glass microchannels, as a function of time (under steady state temperature conditions). The time of appearance of the M-phase shows a strong dependence on the channel geometry (width v/s depth) and surface anchoring (degenerate planar v/s homeotropic), indicating the fundamental role played by the properties of the confining surfaces (specifically, PDMS in this case) that need quantitative discussions detailed in this reference [56]. All our experiments were conducted in multiple replicates (minimum 3) to ensure reproducibility of the results.

Microfluidic confinement
The experimental protocol used in the fabrication of PDMS microfluidic channels is based on the setup described previously [40]. Briefly, microchannels were constructed using polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning) reliefs and prepared by following the standard soft lithography techniques. The substrates were cleaned thoroughly with isopropanol and dried by placing the surfaces on a hot plate (80 • C for 30 min). The molded PDMS reliefs were bonded to glass substrates after being exposed to oxygen plasma. We prepared linear channels with a rectangular cross-section. The channel depth (d) was varied between 10-40 µm, while the channel width (w) was varied between 100-345 µm. The distance between the inlet and the outlet port was set to 20 mm, defining the length( l) of the channel. Different regions along the channel length were observed under microscope to estimate the uniformity of functionalization. At both ends of the microchannel, cylindrical holes with 1.5 mm diameter were punched through the PDMS to provide housing for the tubings. Two teflon (PTFE) tubes with 0.5 mm inner diameter and 1.58 mm outer diameter were inserted into each of housings and served as connectors for inlet-to-source and outlet-to-sink, respectively. For glass devices, hollow rectangle capillaries of dimensions (depth × width): 10 × 100 ± 10 % µm, 40 × 400 ± 10 % µm and 100 × 1000 ± 10 % µm were used to study the texture development of LCLCs. Glass capillaries were either used as such or treated in order to obtain suitable surface anchoring and then filled with the liquid crystal.

Surface anchoring and boundary conditions
Microfluidic devices were functionalized for investigating LCLCs textures within microchannels and glass capillaries possessing two distinct surface anchoring conditions: degenerate planar and homeotropic. Specific anchoring states within the microfluidic devices were achieved by a combination of distinct physical and chemical methods. We applied well-studied plasma exposure (inducing degenerate planar anchoring) and DMOAP treatment (homeotropic) for surface functionalization of the PDMS-glass microchannels and glass-only devices [57,53]. The anchoring conditions were validated using thermotropic nematic 5CB as our control sample, for which detailed studies exist [58].
The degenerate planar surface anchoring is obtained after treatment of the PDMS surface by air plasma. Experiments under degenerate planar alignmnent conditions were conducted under freshly treated channels only. For homeotropic alignment the channels were first filled with 0.4 wt.% aqueous solution of DMOAP and then dried in oven at 80 • C for 2 hours to render both the PDMS and glass surfaces hydrophobic, inducing perpendicular (homeotropic) boundary conditions. In order to achieve homeotropic anchoring of the director at the LC-glass interface, the glass capillaries were filled with a 0.4 wt.% aqueous solution of DMOAP and the solvent was allowed to evaporate before the capillary was filled with the liquid crystal. The anchoring inside the capillaries were characterized through the shape of the meniscus of water and nematic 5CB, relative to the shape inside native capillaries, as shown in Figure 12.

Filling the microfluidic devices
Glass-PDMS and glass capillaries were filled above the isotropic-nematic transition temperature of LCLC solution for a given concentration of DSCG (12, 14 or 16 wt. %) at 45 • C either with a syringe and needle or capillary filling, respectively. Typically, a microchannel kept at 45 • C was filled with a syringe (3ml) with bevelled tip needle (23 gauge, 25 mm long, 0.6 mm outer diameter), to which a PTFE tube (0.5 mm inner diameter and 1.58 mm outer diameter) was connected. The other end of the PTFE tube was inserted into microchannel. After filling, microchannel was sealed with epoxy glue, to avoid flow-induced effects (e.g., alignment) and, crucially, prevent water evaporation that could change the DSCG concentration. Thereafter, microchannels were slowly cooled down to 22 • C with a ramp rate of 1.5 • C per minute. The capillaries were filled with 14 wt. % DSCG solution by capillarity at 45 • C and then immediately sealed with epoxy to prevent water evaporation and to avoid flow-induced effects, and then allowed to slowly cool down to 22 • C with a ramp rate of 1.5 • C per minute.

Optical microscopy and image acquisition
The director orientation of LCLCs in the PDMS-microchannels and glass capillaries was determined using polarised light microscope (Eclipse LV100N-POL, Nikon), equipped with × 5, × 10, × 20 and × 40 objectives and with a Linkam (PE120) heating/cooling stage. The samples were observed between crossed polarizers in a transmission mode using monochromatic light and several polarizer configurations. Additionally, the channels were oriented at 45 • with respect to the polarizers and with and without insertion of a 530 nm phase-retardation plate. The images (6000 × 4000 resolution) and videos (1920 × 1080 resolution) were acquired using using digital a Canon EOS77D camera and videos were recorded in full HD colour at a frame rate of 25 frames per second. The image and video analysis was performed by using tools ImageJ ® and VideoMach ® , respectively.

Scanning electron microscopy
The morphological properties of pristine, oxygen plasma-treated and DMOAP-treated PDMS were analyzed by scanning electron microscopy (SEM). SEM imaging of the PDMS microchannels was done using JEOL JSM-6010LA being operated in 12-15 kV range using an In-lens secondary electron detector. For SEM imaging samples are gold coated (≈1 nm thickness) using a sputter coater (Quorum: Q150 Thin Film Sputter Coater) for 30-120 seconds.

Results and Discussion
The goal of this work is to study the equilibrium textures LCLC under strict microfluidic confinements possessing distinct surface anchoring conditions and channel dimensions. Figure 2 shows the experimental variables considered here: channel dimension, anchoring conditions and experimental timeline. For experiments conducted with PDMS-glass microchannels, all four walls possessed either a degenerate planar anchoring or homeotropic anchoring. From hereon, the results presented will focus on the concentration of DSCG to 14 wt. %, as our control case. Over the following sections, we present the dynamics of the emergence of the LCLC textures due to the DSCG solution. Based on the initial observations, a number of key questions-on the emergence of the dynamic and equilibrium LCLC textures-emerge, answering which could identify the specific roles of surface anchoring, confinement and LCLC concentrations. Below, we present a step by step, detailed report on the emergent textures as a function of the aforesaid factors.

LCLC microfluidic textures due to degenerate planar anchoring
As shown in Figures 2 (a-f), POM imaging was used to characterize the developing director orientation within the microchannels. To create the degenerate planar anchoring, we employed plasma treatment (discussed previously) as a  way to modify the microchannel surfaces to align the nematic phase exhibited by 14 wt. % DSCG. The plasma-treated microchanel was the filled with DSCG, sealed and cooled to 22 • C with a ramp rate of 1.5 • C per minute. In order to perform preliminary optical characterization of the textures, we scanned the entire channel at a lower magnification (5×), followed by images from selected areas (that best represented the overall director profile) were taken at a higher magnification (20×), presented in Figure 2 (a-c). Corresponding textures in homeotropic channels are shown in Figure 2 (d-f), which interestingly, resemble the degenerate planar textures right after the isotropic to nematic phase transition. The anchoring condition was confirmed to be homeotropic by filling in 5CB and comparing the resulting texture with those in our previous works. Figure 3 captures the time evolution of the DSCG textures within channels possessing degenerate planar anchoring. The dimension of the microchannel was kept fixed, with an AR ≈ 10 (w = 100µm, h = 10µm). Panels 1-6 ( Figure 3), show POM micrographs obtained from a combination of different polarizer orientations, used to deduce the average molecular orientations corresponding to the planar textures. The birefringent domains appear shortly after the sample is cooled down below the isotropic-nematic phase transition temperature, and stabilize over a timescale of 60 minutes. This appears as dark and bright regions between crossed polarizers, which show the formation of randomly oriented planar DSCG domains. The textures were further characterized using a λ-retardation plate along with the crossed polarizers (panel 2). It may be worthwhile to note here that the N and M LCLC phases are known to posssess negative birefringence [59], thus, in our POM micrographs with λ-retardation plates, the regions in yellow represent the nematic director orientation parallel to the slow axis of the retardation plate (please see Figure 2 for the reference colour wheel). The region, where the director is oriented perpendicular to the slow axis of the retardation plate appears blue [60]. The colour variation can be attributed to varying LC director orientations of microdomains formed by LCLC stacks in the nematic phase. In order to determine the stability of the planar textures, we checked the texture evolution every 15 minutes after the ramp has finished until 60 minutes. Our observations reveal that initially (right after transition, t=0 minute), the DSCG solution has degenerate planar orientations. Different parts of the microfludic channel were monitored to ascertain the homogeneity of the textures along the channel length. For the region of interest presented here, the planar nematic textures were found to be stable and homogeneous for about 60 minutes.
Interestingly, within similar timescales, over isolated regions within the microfluidic device, we observed the formation of non planar static nematic textures ( Figure 4). Starting from a degenerate alignment, such textures grew and  : Appearance of the herringbone microfluidic texture due to the formation of M-phase in degenerate planar microchannels. Panels 1-6 present the formation of the characteristic herringbone texture (in M-phase) in 14 wt. % DSCG, seen here on the right half of the micrographs. On the left half of the channel, DSCG can be seen to be still in the nematic phase. The micrographs were captured using different orientations of the polarizers relative to the channel, with or without the λ-retardation plate, as described in Figure 3. Here, time (t) is in minutes after the first appearance of the nematic domains. Microchannels were 100µm wide and 10µm deep, scale bar: 50 µm.
coated LC glass cell [15]. It was also observed that degenerate planar texture evolved to ribbon structure, texture corresponding to coexistence of M and isotropic phase. As there is no preferential alignment of the director along any direction, the DSCG stacks orient randomly forming a multi-domain structures of different orientations of the director. However, the timescale over which the planar textures transition from the nematic to herringbone (M-phase) depends on the channel aspect ratio and the DSCG concentration, as discussed in Section 3.3.

Nematic texture in homeotropic microchannels
We explored surface treatment with silanes as a way to modify the surface of PDMS-glass microchannels that support homeotropic anchoring. DMOAP-treated microchannels (prepared freshly) were filled with 14 wt. % DSCG solution in isotropic phase, sealed and then allowed to cool down to 22 • C, maintaining a steady ramp rate of 1.5 • C per minute, as reported previously for the plasma-treated glass-PDMS microchannels. Figure 5 summarizes the resulting LCLC textures produced by 14 wt. % DSCG solution within homeotropic PDMS-glass microchannels. The non-zero birefringence patterns suggest that the DSCG solution forms organize into unformly aligned planar or tilted domains within PDMS-glass microchannels, over the observed time window.
By capturing the evolution of the textures over time, we observe that, initially, the LCLC nematic textures resemble the degenerate planar textures, even under homeotropic boundary conditions. We call this as the pseudo planar texture, arising due to planar or tilted organizatin of the DSCG nematic discotic stacks. This is in contrast to the microscale thermotropic nematic textures formed due to homeotropic boundary conditions: here the polarized light undergoes complete extinction upon passing through the sample (thus, appearing completely dark).
As for the case of the degenerate planar anchoring (previous Section), for the homeotropic boudaries too, we observe isolated areas along the channel where the M-phase nucleates, and thereafter propagates through the channel, filling it up completely ( Figure 6). However, unlike the degenerate planar case, the M-phase texture under the homeotropic boundary conditions manifests as the characteristic spherulite texture. These contrasting outcomes-herringbone microfluidic texture in degenerate planar, and spherulite texture in homeotropic boundary conditions-suggest that the anchoring conditions play a key role in establishing the stable equilibrium static LCLC textures in PDMS-glass microfluidic confinements. Nazarenko and co-workers showed a weak homeotropic anchoring on hydrophobic silanetreated substrates [23], which eventually at a later time, transitioned from homeotropic to planar anchoring, while Tone et al. reported that use of silane to get homeotropic alignmnet does not work on glass LC cells [24]. From the texture evolution observed in Figure 5 it can be concluded that for the tested silane (DMOAP)-treated PDMS-glass microchannel the nematic texture of DSCG does not show the typical homeotropic alignment and furthermore, the areas with homeotropic textures were found to be unstable. A time lapse sequence of the spontaneous transition from the pseudo planar to spherulite texture is presented in Figure 34.

Role of the channel dimensions on the evolution of the LCLC textures
The role of microfluidic dimensions was investigated by varying the channel aspect ratio, the ratio between the channel width w and height h (AR = w/h). We have studied the texture evolution by systematically varying both the channel width and height, through the following combinations of w x h (all dimensions in micrometers): (i) Variation of width (in µm) -100 x 10, 200 x 10 and 345 x 10; and variation of height (in µm) -100 x 10, 100 x 25 and 100 x 40.

Role of the channel width
The role of channel width was investigated by studying two additional dimensions: 200µm and 345µm, while keeping the height fixed at 10µm. All other experimental conditions and parameters were kept similar to those used for the experiments reported in the previous Sections (with 100µm channel width). As described earlier, the stability of the degenerate planar anchoring was checked every 15 minutes after the cooling ramp concluded (this coincided with the first appearance of the birefringent domains). For both the 200µm and 345µm wide microchannels, the observations lasted up to 60 minutes (or slightly longer, as the case was in a few instances), till the equilibrium LCLC static texture was observed, i.e., no further change in the overall pattern and birefringence. The results are summarized in Figure 7 using POM micrographs of the microchannels (200µm wide, 10µm deep) under different crossed polarizer orientations.
Initially, we observe the expected planar texture of DSCG due to the degenerate planar anchoring conditions on the channel walls.The planar microdomains, signifying different orientations, gave way to nucleation of the M-phase herringbone texture, wherein the discotic columns pack in more ordered two-dimensional lattice. Starting from a degenerate alignment (Figure 7), the herringbone texture proliferated along the channel length, and over a timescale of about 60 minutes, covered the entire channel (shown in Figure 8). It might be worth noting that the herringbone texture manifests a global orientation parallel to the length of the microchannel: this suggests that-under degenerate planar anchoring conditions-the microfluidic geometry underpins the emergent textural anisotropy of the confined DSCG (see Figure 29). On increasing the width to 345 µm, as shown in Figures 13 and 14 we observed the textural transition from schlieren to herringbone texture consistently, however for wider channels the transition occurs over a shorter timescale. The relation between the transition time and channel aspect ratio will be reported elsewhere [56]. In summary, the degenerate planar anchoring evolves into M-phase well-aligned herringbone texture over a time span of 60 minutes, such that wider the channels, longer is the transition time (positively correlated).
Moving from degenerate planar anchoring to homeotropic anchoring conditions, initially the characteristic homeotropic textures were observed (complete extinction of the polarized light) within the 200 µm wide, 10µm deep DMOAP-treated microchannels. Over a timescale of 15 minutes, the homeotropic domains gradually transform into the pseudo planar texture, as shown in Figure 15. Finally the planar texture transforms into the M-phase, over a timescale of 30 minutes. Apart from textures observed in area of microchannel as shown in Figure 15, two other areas of this microchannel were observed: one where where M-phase ( Figure 16) and other with homeotropic texture (Figure 17), developed at t=45 minutes. For areas where homeotropic texture developed, on observing between crossed polarizers and rotation of sample by 45 • , intensity of transmission light remained low. This area with homeotropic alignmnet further propagated in the microchannel until t= 60 minutes but eventually changed to a degenerate planar texture. Another part of the microchannel where M phase appeared at t=30 minutes ( Figure 6), after 45 minutes the observed area completely changed to M-phase. On changing the channel width further to 345 µm (Figures 18 and 19), qualitatively, we observed the same trend: regions of low intensity of transmitted light were first observed, which then transformed into the pseudo planar (within t= 15 minutes), which eventually changed to the M-phase at t= 45 minutes. It may be worthwhile to repeat here that, under homeotropic boundary conditions, the M-phase is distinct from that within degenerate planar microchannels: In hometropic channel, the spherulite texture-stretching transversely across the channel length-is observed, whereas within the degenerate planar channel, the herringbone structure organizes along the channel length.
Taken together, these observations indicate that the channel geometry and surface anchoring play a key role in the development of the static LCLC textures. Overall, the change in the channel width impacts the nematic to M-phase transition time scale: an increase of 100 % channel width (from 100 µm to 200 µm) lead to the time of appearance of M-phase increase from by 100 % and 200 % for plasma and DMOAP-treated microchannels, respectively. On further increasing the width by 245 % (from 100 µm to 345 µm) the time of appearance of M-phase is increased by 200 % for both plasma and DMOAP-treated microchannels. These results motivated us to investigate the effects of the change of the channel depth, while keeping the channel width constant at 100 µm and 200 µm.

Role of the channel height
The effect of the channel height was studied using microchannels with h = 25 µm and 40 µm, while maintaining the channel width w = 100 µm. The time evolution of the DSCG textures obtained for plasma-treated microchannels of dimension 25µm and 40µm via POM are shown in Figure 20-21 and Figures 22 and 23, respectively. On increasing the depth by 150 % (i.e., from 10 µm to 25 µm deep), the M-phase appeared between 45-60 minutes, whereas on increasing the depth by 300 % (100 µm wide, from 10 µm to 40 µm deep), the M-Phase appeared between 60-75 minutes. So, the appearance of the M-phase, and hence, the nematic to M-phase transition, is retarded due to the increase in the  Figures 24 and 25, respectively. Here too, the transition time scale is stretched, relative to our control case: the M-phase appears at about 60 minutes within the 100µm wide, 25µm deep channel; while no phase change was captured within the first 60 minutes inside the 100 µm wide, 40 µm deep channel. We also studied time evolution of textures for channels with 15µm depth and 200µm width, both for plasma (Figure 26-27) and DMOAP ( Figure 28) treatment. In general, we observed that upon increment of the channel depth from 10 to 40 µm, the DSCG texture changes from uniform to non-uniform appearance, for both the degenerate ( Figure 22) and the homeotropic (Figure 25) boundary conditions. Similar observations on texture uniformity were made when channel depth was changed from 10µm to 15µm for 200µm wide channel, as shown in Figure 26 and Figure 28.
Based on the above results, we conclude that the stability of the nematic and M-phase textures, and the transitions therein, depend closely on the channel dimensions and the surface anchoring conditions. Furthermore, the height of the microchannel regulates the uniformity of the emergent textures: lower the height (i.e., stronger confinement) leads to uniform texture, whereas in deeper channels the textures become increasingly non-uniform. Additionally, we also used two other DSCG concentrations for our experiments: 12 wt. % (closer to the isotropic phase at the operating temperature) and 16 wt. % (closer to M-phase) and studied the POM textures 200 µm wide and 10 µm microchannels. As shown in Figure 9 (panels 1-4), initially no microfluidic texture was detected at 12 wt. % DSCG, under both degenerate planar and homeotropic boundary conditions. Then, between 45-60 minutes, planar nematic textures emerged, which at about 90 minutes changed to the M-phase texture. At 16 wt. %, within DMOAP-treated microchannel (panels 5-6), a patchy homeotropic domain was observed (with regions of total light extinction), while rest of the regions were birefringent, which ultimately transformed to M-phase at around 60 minutes. Figure 10 and Figure 29 shows the M-phase texture developed in PDMS-glass microchannels, under both plasma (degenerate planar) and DMOAP (homeotropic) surface treatments. For the microchannels with plasma treatment, the M-phase yields the herringbone texture, with an overall structural orientation along the channel lenght. For the DMOAP-treated microchannel, the M-phase elicits the spherulite texture, which aligns perpendicular to channel walls, i.e., along the transverse direction ( Figure 10 and Figure 29). As indicated by contact angle measurement of 14 wt. % DSCG solution (Figure 1, panels f and g), plasma exposure renders both PDMS and glass surfaces hyrophilic, whereas the silane, DMOAP, makes the surface hydrophobic. Consequently, nucleation, growth and propagation of the textures will be locally influenced by the immediate boundary conditions. In the planar case, the discotic stacks in the nematic or M-phase are expected to align along the planar domains imposed by the plasma treatment. This, with the additional constraint from the long axis of the microchannel, guides the herringbone structure along the channel length. For the homeotropic case, the spherulite texture emerges due to normal alignment of the discotics on the channel surfaces, and are thus constrained to grow from ceiling to the floor, and from wall-to-wall of the microchannel. Thus, we do not see any structural bias along the channel length. In order to gain a mechanistic insight, we looked into impact of oxygen plasma and DMOAP treatments on the PDMS surface morphology. Figure 30 shows topography of the pristine, oxygen plasma-treated and DMOAP-treated PDMS. It has been known that plasma treatment of PDMS surface causes nano-structuring of PDMS and increase water contact angles depending on duration and intensity of plasma treatment used [61,62,63,64]. Bodas et al. showed the notable cracks (through SEM) to the PDMS upon 150 W, 15 min, plasma treatment [61]. On the other hand, Hillborg et al. employed 40 W, 3 min, plasma treatment and observed insignificant surface morphological changes (through SEM and AFM) [64]. From our SEM images, no significant change is visible between the untreated and plasma-treated PDMS; however the DMOAP-treated PDMS shows a non-uniform patterned surface. On looking mainly from the surface hydrophilicity/hydrophobicity angle (which is a robust and reliable outcome of oxygen plasma treatments on PDMS), Ruben et al. had shown the enhanced capillary flow within PDMS microchannels that are hydrophilic [65]. Looking from a membrane point of view, hydrophobic species can permeate readily through PDMS, ranging from pharmaceutical compounds to organic dyes [66,67] but it is highly size dependant. Since, DSCG is a bulkier molecule with much higher solubility value (195.3 mg/ml [68]) compared to small molecules like Quinine or Nile red, we can rule out the possibility of DSCG permeation in PDMS. Nonetheless, in relation to DSCG, the absence of birefringent patterns in bulk PDMS is an additional support that DSCG does not permeate into the PDMS bulk. It can be concluded that a more hydrophilic PDMS surface would enhance the permeation of water through PDMS.

Effect of surface treatment on LCLC microfluidic textures
It is evident from the time-dependent optical observations that the nematic phase of 14 wt. % DSCG, within both plasma and DMOAP-treated microchannels, exhibited a phase transition from the nematic to the more ordered columnar M-phase. It is known that M-phase occurs at higher concentration of DSCG and for an N-M phase transition to occur from a nematic phase, there has to be change in concentration of DSCG (unless the temperature of the environment changes, which in our experiments, was kept fixed). The change in DSCG concentration can happen either on changing temperature or loss of water. As all the measurements were done at a constant temperature, the change in concentration of initial DSCG solution has to come from loss of water. In order to ensure that the change in textures of nematic phase to M-phase is a result of water loss due to the PDMS-glass microchannels, we compared the temporal stability of the DSCG textures (for 14 wt. % DSCG solution) within glass capillaries, for both native and DMOAP-treated samples. As shown in Figures 31 and 32, for untreated and DMOAP-treated glass capillaries respectively, the textures for planar and homeotropic alignment were observed. The time dependant texture analysis showed no N-M phase transition, regardless of the surface treatment. We also studied nematic phase of 14 wt. % DSCG solution by keeping one end of glass capillary open to allow water evaporation. Although the time evolution of texture showed a texture with birefriengence change, no herringbone M-phase texture was observed. This is shown in Figure 33. The above combination of control tests confirms that the M-phase development within the PDMS-glass microchannels is mediated by the surface attributes of PDMS. We studied 12 wt. % DSCG solution in a plasma-treated PDMS-glass microchannel Figure 34. As this concentration is close to the isotropic-nematic phase traition temperatire (see, phase diagram in Figure 1), one should expect that the system has to first reach a concentration of DSCG that shows a stable nematic phase, and then the N-M phase transition follows. We filled a plasma-treated PDMS-glass microchannel with 12 wt. % solution and sealed it, and as discussed previously, carried out POM time lapse imaging (imaged every 30 seconds). As hypothesized, initially the system was observed to be in the isotropic phase, with the nematic phase emerging after about 25 minutes. Thereafter, the system stays in nematic phase for about 255 minutes, before the M-phase finally appears. This confirmatory test consolidates our conclusions that the surface properties of PDMS induces systematic changes in the initial concentration of DSCG, which ultimately lead to the formation of the spontaneous, yet tunable, LCLC phase transitions. A detailed quantitative analysis of the surface-mediated phase transition will be discussed elsewhere [56].

Conclusions
We exploit a simple and flexible soft lithography technique to create PDMS-glass microfluidic devices for studying the nucleation, growth and stability of LCLC textures. The temporal study carried out here reveals spontaneous (and unexpected) emergence of M-phase textures, with DSCG solutions which are originally in nematic phase. The comprehensive polarization optical study has demonstrated that static LCLC microfluidic textures are far from trivial, and are finely tuned by three key parameters: (i) microchannel dimensions (confinement effect), (ii) surface anchoring (anchoring effect), and crucially (iii) experimental timescale (spontaneous surface-mediated effects) The three fundamental variables determine the local and temporal no-f low LCLC microfluidic textures, giving us a set of key experimental parameters that could be potentially employed to register the initial conditions for microfluidic f low experiments.  homeotropic alignment, for instance polybutadiene and styrene-butadiene-styrene [24]. These materials can work for glass devices, however there are a few technical hurdles in coating PDMS-glass microchannels with these materials, primarily due to the dissolution step (with organic solvents) which can cause swelling of the PDMS, thereby modifying the microfluidic geometry.
The key and the most striking result of this work is the temporal variation of lyotropic textures induced by a spontaneous phase transition from nematic to the M-phase. By changing the channel dimensions, surface anchoring, and the concentration of the LCLC material, we could systematically tune both the nature of the characteristic texture and the temporal dynamics of the nematic-to M-phase transition. Additionally, the texture of the M-phase depended on the surface anchoring conditions: herringbone texture under degenerate planar conditions; and spherulite texture under homeotropic conditions. Furthermore, both herringbone and spherulite textures manifested global orientations: while the herringbone texture was oriented along the channel length, the spherulite texture was oriented perpendicular to the channel length (in the transverse direction). The temporal dynamics of the nematic-M-phase transition could be controlled by varying the channel width (larger width lead to longer transition time), whereas changing the depth primarily impacted the textural uniformity. Our straightforward strategy to employ LCLC self-organization within softlithography based microfluidic platform considerably widens the scope of LCLCs, for both applied and fundamental research, and sets the stage for future studies on the flow behaviour LCLCs in strict microfluidic environments. Interestingly, even the static no-f low textures hold promising prospects for potential applications in LCLC-mediated transport processes. As shown through time lapse image sequence in Figure11, defect lines near a microscale inclusion (channel and inclusion were DMOAP-treated) can transition over time-purely due to surface-mediated effects-thereby triggering microscale transport, even in absence of external pressure gradients [56]. Investigations on the dynamics of LCLCs under flow within PDMS-glass microfluidic devices is currently underway, and can be extended to other systems to study surface-mediated phase transitions, for instance in water based lyotropic liquid crystals systems such as peptides, liquid crystal polymers, and viruses. The temporal stability of textures can be of significant implication on real-world applications, particularly to capture the dynamics of in biological systems, where transport processes can respond to (or be modulated by) the compliance, surface and morphological attributes of the system.

Author contributions
A. Sharma and A. Sengupta designed research, performed experiments and analyzed data. All authors interpreted the data and wrote the paper.  Here, time (t) is in minutes after the cooling ramp is completed). Scale bar: 50 µm. Figure 14: Herringbone texture in due to degenerate planar anchoring within 345µm wide, 10µm deep microchannel. The POM images were taken using the same protocol as described above. Scale bar: 50 µm