Colloidal Particles in Blue Phase Liquid Crystals

Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains. Accepted Manuscript


Introduction
Colloid-liquid crystal composites are an exciting class of responsive, soft materials. Colloidal particles mixed into liquid crystals create defects in the (ideally defect free) ordered phase. The form of the defects is dependent on the particle size, the alignment of the mesogens at particle surface and how strongly this alignment is enforced (the anchoring strength). Colloids tend aggregate in order to share defects and minimise the disruption of the ordered phase. Structures including lines 1 , plates 2 and 3d colloidal glasses 3 are formed. An even richer range of colloidal structures are observed when the liquid crystal is chiral 4,5 .
One such chiral liquid crystal phase is the cholesteric blue phase. The blue phases of chiral liquid crystals consist of an ordered array of defect lines. They only exist in a narrow range of temperatures and chiralities between the isotropic and cholesteric phase due the delicate balance between satisfying an increased degree of twist at the expense of the formation of defect lines. Depending on temperature and chirality, these defect lines form either a cubic lattice in blue phase one (BPI) and blue phase two (BPII) and an amorphous network in blue phase three (BPIII) 6 .
Simulation work 7,8 has suggested that the cubic structure of the blue phase lattice could be used as a template to assemble 3d photonic crystals and other switchable opto-electronic structures. In particular varying the size of colloids in composite blue phase-colloidal particle photonic crystals is predicted to alter the photonic band gap 8 . However, the addition of particles to a BP raises the question of how the particles will interact with a phase already dominated by defects. particles with weak surface anchoring such as those used by Cordonyiannis et al. 9 the particles are assumed to contribute to the stability of the blue phase by occupying the defect cores of an undistorted blue phase lattice as in simulation 10 . For particles much larger than the defect lines, it is clearly essential to be able to visualise what happens close to the particles' surfaces.
When particle size and anchoring strength are considered a rich variety of colloidal structures are observed, dependent on the anchoring strength at the colloid surface. Recent simulation results 11 have shown that the occupation of an undistorted lattice is an ideal case. For weak anchoring at the particle surface, the blue phase lattice acts as a template for the colloids, but as the anchoring strength increases the colloids' own defects become dominant and the blue phase lattice is distorted.
There have been only limited experiments on micron-sized colloidal particles in blue phases 12 . These studies have focused on the ability of colloidal particle to increase the (temperature) existence range of the blue phase. They show that for micron-sized colloidal particles there is only a negligible increase in blue phase stability. There are no published images showing the changes to the ordered state due to colloids dispersed within the blue phase. In this study we use confocal microscopy to observe large (r = 1.5 µm) colloidal particles dispersed in a cholesteric blue phase in order to explore the influence of the colloids on the behaviour of the blue phases.
We build on previous studies of these colloidal particles dispersed within cholesteric and nematic liquid crystals both in the bulk 13 and at interfaces 14,15 . The colloids have strong planar anchoring at their surface and form flat plates perpendicular to the helical axis in the cholesteric phase.
In order to prepare a blue phase we take liquid crystal through a phase transition with the colloids already dispersed. Colloidal particles introduce disorder, which has a significant effect on phase transitions 16 . It has been shown that the addi-1-9 | 1 tion of quenched disorder effectively smooths out a first order phase transition, making it second order above a certain degree of randomness 17 . The theory of disorder in liquid crystals has been applied successfully to the case of aerosils dispersed in nematics 18,19 , smectics 20,21 and nematics confined within in random pore glasses 22 . An extensive review can be found in Popa-Nita et al. 23 . Further, Cordonyiannis et al. 9 report that the particles affect the nature of phase transitions, in particular the transitions from the isotropic phase to the BP. The transitions become broader and "smear out" with increased concentrations of nano-particles.
The effect of disorder on phase transitions from more ordered to less ordered phases is pertinent to systems beyond the liquid crystal colloid composite reported here. In the field of nuclear reactor materials, the presence of a disordered phase between an ordered phase and the melt has been shown to confer increased stability and resistance to radiation damage 24 . The melting behaviour of inclusions in metallic systems is of interest for engineering applications and is often difficult to observe in-situ 25 .
Below, we explore how a system already dominated by defects responds to disorder. We observe micron-sized colloidal particles dispersed in a blue phase exhibiting LC. The effect of particles on the behaviour of the liquid crystal on heating from the cholesteric phase through the blue phase to the isotropic is explored. We observe that well ordered BPI and BPII form in the presence of the colloids, but that subsequently localised melting occurs.

Mixture
Concentration of chiral dopant Pitch length Long pitch 7% 0.40 µm ± 0.01 µm Short pitch 10% 0.29 µm ± 0.01 µm The liquid crystal is a three-component commercial mixture. It consists of two nematic liquid crystals, 5CB (Aldrich) and a fluorinated nematic mixture, JC1041-XX (Chisso). Finally a chiral dopant, ZLI-4572 (Merck) is added to create a chiral nematic liquid crystal. Varying the concentration of the chiral dopant varies the pitch of the liquid crystal. The composition and estimated pitch length of the liquid crystals used in these studies can be found in table 1 This was calculated using values for the helical twisting power of ZLI-4572 given by Yan et al. 26 . Short pitch mixtures exhibit both BPI and BPII. Long pitch mixtures exhibit only BPI. The three components were mixed by stirring overnight until uniform.
Colloidal particles, fluorescein isothiocyanate (FITC)labelled melamine with carboxylate-modified surfaces pur-chased from Fluka, r = 1.5 µm. These were dried under vacuum at 40 • C overnight prior to use. The colloidal particles were dispersed in the cholesteric phase of the liquid crystal mixture using stirring, sonication (via an ultrasound bath VWR Ultrasonic Cleaner) and vortex mixing. Care was taken to ensure that the liquid crystal did not enter the isotropic phase. Cooling from the isotropic in the presence of particles can result in the formation of a cellular solid 27 . Once the samples were uniform (confirmed via visual inspection and microscopy) they were used immediately. If samples were stored then the colloidal particles were re-dispersed prior to use. This dispersion method was chosen over a co-solvent transfer method to avoid the risk of contamination. The final volume fraction was 3 %.
To explore the effect of particle size a some experiments were performed with r = 0.5 µm (FITC)-labelled melamine, carboxylate-modified colloids also from Fluka.
Once the colloidal particles were uniformly dispersed samples were loaded into glass capillaries (0.05 x 1 x 50 mm) or hand built sample cells via capillary action. The cells were sealed using silicone vacuum grease (Dow Corning) and used immediately.
Blue phases are only prepared by heating from the cholesteric as the colloidal particles rapidly sediment out from the isotropic phase. When blue phases are prepared by heating, the elastic nature of the cholesteric and blue phases prevents the colloidal particles from sedimenting. Bulk samples of colloidal particles in the cholesteric phase are stable to sedimentation for months.
Some preliminary experiments were performed using an Instec Hotstage in conjunction with the confocal microscope to allow the colloidal particles to be imaged in three dimensions.
To allow a greater control of heating rate a custom heating stage was constructed. The temperature is controlled via a Lakeshore 331 temperature controller to ± 0.005 • C. Samples were mounted in the hotstage using heat sink paste (RM) to ensure good thermal contact. A lid was placed over the sample inside the stage body to increase thermal stability. Once mounted, the sample was heated rapidly (10 • C min −1 ) to within 5 • C of the blue phase transition. The heating rate was then decreased to 1 • C min −1 . For the final stage (within approximately 1 • C of the transition) the heating rate was decreased further to 0.1 • C min −1 . During heating, time lapse images were recorded using a Zeiss Observer.Z1 inverted confocal microscope in conjunction with a Zeiss LSM 700 scanning system and a 20x NA=0.8 objective. The FITC in the colloids was excited using 488nm diode laser. The transmission signal from laser was also recorded. Once the blue phase transition was observed the temperature was kept constant until the transition was complete. Further heating and cooling of the sample was then performed within the blue phase to study the effect of the colloidal particles. The temperature 2 | 1-9

Soft Matter Accepted Manuscript
was recorded digitally via an adapted LabView application provided by Lakeshore.

Colloidal Particles in the Cholesteric Phase
Colloidal particles mixed into cholesteric liquid crystals have been shown to stabilise the oily streak network forming a colloidal particle stabilised defect gel 5 . Colloidal particles with planar alignment form flat sheets perpendicular to the helical axis 13 . In this study similar behaviour is observed; flat plates of colloidal particles are found surrounded by defects and always at the junction of oily streaks as can be seen in fig. 1. On heating towards the blue phase transition the oily streak network undergoes significant rearrangement. This rearrange-ment of defects results in some movement of the associated colloidal particles. However, on average, the colloidal particles remain in flat plates perpendicular to the helical axis and surrounded by defect lines. This was confirmed using confocal microscopy in conjunction with the Instec stage. Fig. 2 The transition from cholesteric to blue phase in a planar cell. These images were taken using a confocal microscope between crossed polarisers. False colour has been used to enhance contrast.

In the absence of colloids
In the absence of colloids the oily streak network and other defects present in the cholesteric act as nucleation sites for the blue phase. In fig. 2 a sample with a dense oily streak network and no colloidal particles is heated rapidly (5 • C min −1 and then 1 • C min −1 ) to close to the blue phase transition temperature. The rapid heating ensures that the defect network does not have time to anneal. On heating through the transition the oily streak defects thicken as the blue phase grows from the cores fig. 2(b) and 2(c). The blue phase grows outwards from the defects with islands of aligned cholesteric texture fig. 2(d) the last to disappear. The observation that the blue phase grows preferentially from preexisting defects, either in the form of oily streaks or defects surrounding colloidal particles is unsurprising. The formation of BPI from a uniformly aligned cholesteric requires the formation of a lattice of defects. When the cholesteric already possesses defects it is likely that there is 1-9 | 3 a lower cost for defect formation and the blue phase can preferentially grow from the defect regions. The blue phase which grows from these regions exhibits a range of colours; indicating different alignments of the ordered BPI grains as in fig.  4. This is in contrast to the BPI in regions where the parent cholesteric had uniform alignment. In the aligned regions the blue phase has a single colour indicating that the grains have similar alignments.
The transition from cholesteric to blue phase takes around a minute to complete. We measure the speed at which the blue phase expands into the aligned cholesteric manually using im-ageJ. The phase fronts which advance from the defects into the aligned cholesteric move at 0.5 ± 0.1 µm s −1 . This is slow compared to velocities reported for nematic isotropic fronts geqslant100 µm s −1 28 reflecting the highly visco-elastic nature of the cholesteric and blue phases. During these transitions the temperature changes at a maximum rate of 0.1 • C min −1 and by at most 0.1 • C (due to lag in the temperature stage). Even if the temperature stage is set to hold at a given temperature once the transition is under way it continues to completion. We do not see long lasting cholesteric blue phase coexistence.
Simulation studies by Henrich et al. 29 predict that homogeneous nucleation of a BP from a uniformly aligned cholesteric forms a disordered blue phase lattice similar in nature to BPIII rather than the equilibrium structure of either BPI or BPII. These disordered structures are kinetically stabilised. Experimentally, as seen here, it appears that hetrogeneous nucleation from defects is more likely and these disordered structures are avoided.

In the presence of colloids
In the presence of colloids the particles stabilise the oily streak network 5 pinning it in place. Colloidal particles are themselves surrounded by defects in the cholesteric phase 13 as can be seen in figures 1 and 3(a). All of these defects act as nucleation sites for the blue phases. which can be seen to grow from the colloidal particles and the oily streaks simultaneously, fig. 3(b). As in the case with no dispersed colloidal particles, the islands of aligned cholesteric between defects are the last to transition to the blue phase fig. 3(c). The particles have little influence on the cholesteric to blue phase transition, other than increasing the stability of the oily streak network which acts as a nucleation site from the blue phase. The velocity of the phase fronts is unaffected by the presence of colloids.

Blue Phase to Isotropic Transition
Adding colloidal particles to a blue phase liquid crystal has a clear effect on the kinetics of the phase transition between the blue phase and the isotropic. They also lead to a shift in the transition temperature.

Low chirality samples
As discussed in the previous section, colloidal particles and their associated defects act as nucleation sites for the blue phase. In low chirality samples once the sample is in BPI the colloidal particles are observed surrounded by uniform BPI, fig. 3(d). On further heating but at temperatures below the final melting point dark regions are observed around the colloidal particles. Colloidal particles within these regions move due to Brownian motion indicating that locally the elastic modulus is lower. Very occasionally dark regions without colloidal particles can be observed once the temperature is raised further. This is rare, typically blue phases melt from the regions around colloidal particles which grow to fill the whole sample. Finally, the whole blue phase melts. The dark regions become completely black. This indicates that the regions are present though the full height of the sample. Colloidal particles within these black regions sediment through the isotropic phase onto the sample boundaries.

High chirality samples
In samples with higher chirality the liquid crystal can exhibit both BPI and at slightly higher temperatures BPII. The pres-   ence of colloidal particles has no discernible effect on the BPI to BPII phase transition, fig. 5. Unlike the cholesteric to BPI transition there is no evidence of nucleation from the colloidal particles or BPI grain boundaries. Once the sample is heated further in BPII, dark regions similar to those seen in lower chirality samples surround the colloidal particles, fig. 5(d). The transition temperatures of these phenomena with reference to the cholesteric to BPI transition can be found in table 2. We note, figs. 5 and 7 give an indication of BPI and BPII grain sizes, which are unaffected by the presence of colloids.

The phase of the inclusions
Once the melted regions have appeared colloidal particles are free to move within them. Colloidal particles do not leave the dark regions. If dark regions shrink, due to a temperature fluctuation, clusters of colloidal particles become more compact indicating a phase separation. There are two candidate phases which exist at higher temperatures than BPI and BPII which have reduced birefringence. These are BPIII and the isotropic BPIII, an isotropic disordered lattice of defects 30 , is only weakly birefringent and appears as a blue fog in polarising optical images. The materials used in our study do not exhibit BPIII in the absence of colloidal particles. However, given that nano-particles have been shown to stabilise BPIII 31 and disordered blue phases have been shown to appear at phase transitions 29 and in the presence of colloidal particles 11 , a disordered blue phase cannot immediately be discounted as a candidate for the dark regions.
A more compelling reason to conclude that the dark regions are composed of the isotropic phase is that the colloidal particles move in these regions. BPI and BPII are visco-elastic in character 32,33 . The cholesteric phase and BPIII are highly 1-9 | 5 viscous and possess a similar viscosity to one another (η= 0.1 Pa s) 34 . The isotropic phase is a Newtonian fluid. Using particle tracking routines in IDL developed by E. Weeks and others 35 we measure the mean squared displacement of the of colloidal particles within these dark regions and extract a diffusion constant. From this we calculated associated viscosity of the melted regions to be η = 0.005 ± 0.002Pa s. This is the same order of magnitude as that of 5CB at the same temperature in the isotropic phase, 0.016 Pa s implying that the dark regions are inclusions of the isotropic phase. The final value for the diffusion coefficient extracted from the particle tracking results must be used cautiously as the equations used to extract the diffusion coefficient are not applicable to partially ordered fluids. However, the fact that the colloids move at all in these regions and are stationary in both the cholesteric and blue phases is highly suggestive that the region is in the isotropic phase given the difference in viscosity between the LC and isotropic phases.
The presence of the isotropic phase close to the colloidal particles suggests that colloidal particles are acting as nucleation sites for the isotropic phase. There is only limited work on effect of large colloidal particles on an ordered blue phase. Simulations of small colloidal particles with weak or zero surface anchoring dispersed in blue phases find that they are localised at the defect junctions 7 of a cubic blue phase. However, this arrangement is extremely sensitive to the anchoring strength and size of the colloidal particles. For even moderate adjustments the blue phase defect structure becomes disordered 11 . Ravnik et al. 36 perform simulations on micron size colloidal particles in confinement. They find that for finite anchoring strength at the colloidal particle surface the colloidal particles are surrounded by a cage of defects. Table 2 shows the transition temperatures of the low and high chirality blue phase with colloids with reference to the cholesteric to blue phase transition temperature. There is no significant difference in the range of existence of the blue phase with and without colloids.

Transition Temperatures
We also record the appearance temperature of dark regions which appear below the bulk blue phase to isotropic transition with reference to the (bulk) blue phase to isotropic temperature. It should be stressed that these regions appear below the bulk transition temperature and remain local to the colloidal particles. They do not grow further if the temperature is kept constant. The temperature must be raised in order for the bulk blue phase to melt. The vast majority of these regions are observed to appear close to colloidal particles, however a small number are observed without colloidal particles. No such regions are observed in samples without colloidal particles.
The presence of the dark regions suggests that we have entered a biphasic region of the phase diagram, with a colloid rich disordered phase coexisting with a colloid poor blue phase. We do not see coexistence between the blue phase and isotropic on heating in the absence of colloids. However, the blue phase to isotropic transition is first order and coexistence between the blue phase and isotropic has been observed in calorimetry 37 and in microscopy on cooling from the isotropic 38 . This would suggest that the colloidal particles increase the temperature range of coexistence between the blue phase and the isotropic. Studies of disorder in nematic systems have shown that the presence of disorder lowers the nematic -isotropic transition temperature. This appears to be the case here, as the temperature of the cholesteric to blue phase and the blue phase to isotropic transitions are lower close to colloidal particles which act as regions of disorder.
The fact that the colloidal particles expand phase coexistence between the blue phase and the isotropic might lead us to expect that they would also promote coexistence between BPI and BPII. However, we do not observe such coexistence. We propose the a tentative explanation; the region around the colloidal particles will have a higher concentration of defects than the rest of the sample regardless of the nature of blue phase. These defects do not have the structure of either blue phase and so present no advantage for the nucleation of the higher temperature ordered phase. On a more fundamental level it suggests to us that the picture of disorder enhancing phase coexistence at first order phase transitions 17 is unhelpful in the BPI to BPII case. The faceted nature of the melted regions as shown in fig. 6 shows that they are constrained by the elastic anisotropy of the embedding blue phase. Faceted regions are larger and more clearly visible in BPII. In BPII the grain size of the blue phase is significantly larger than the faceted regions. Facets within a grain have the same orientation. In BPI the grain size is typi-

Facets
blue phase to the isotropic at temperatures below the bulk BP-isotropic transition temperature. The shapes of the locally melted regions are anisotropic, suggesting the anisotropic nature of the blue phase-isotropic surface tension influences the shape of the melted regions. In contrast to particles dispersed in a cholesteric liquid crystal, where colloids are held within a system spanning oily streak network, 5 this melting is a purely local phenomenon: it is unaffected by the BP grain size. Compared to the cholesteric, there are no associated non local defects, equivalent to the oily streaks. Colloidal particles have previously been shown to create defects in ordered chiral phases 7,41-43 . It is possible that for the cholesteric to blue phase transition these defects act as nucleation sites for the higher temperature, defect dominated blue phase. Within the blue phases, recent simulation work 11 , has shown that particles with strong surface anchoring disrupt the blue phase lattice, this disruption may favour melting to the isotropic. This simulation is with particles which are smaller than the BP unit cell. For larger particles, a cage of disordered defects is observed 7 . In both cases the disorder is local to the particle as observed in our experiments.
The area of the melted region per colloid is independent of the size of the colloidal cluster, we conclude that the energetics is not strongly influenced by the particle surfaces. Instead this suggests that the area of the melted region is controlled by the anisotropic blue phase-isotropic surface tension rather than by a simple heterogeneous nucleation mechanism. There is no difference in the melted area per colloids between inclusions in the two blue phases. These behaviours are in marked contrast to research where nano particles extend the range of these subtle phases.