The adsorption of fluorinated dopants at the surface of 5CB: a neutron reflection study

The adsorption of dopants at the surface of 5CB has been studied using neutron reflection. The dopants were versions of 11OCB with partly fluorinated chains, and the 5CB was interfaced with air or with silica treated with fluorocarbon or hydrocarbon coatings. At the air interface, the F17-11OCB adsorbed homeotropically, and the amount increased on cooling into the nematic phase of 5CB. At low temperatures, the adsorption from the nematic phase appeared to saturate at a bilayer. At the solid interfaces, there was some evidence for a thin planar layer at the surface, but most of the adsorbed dopant was in the form of diffuse layers. In both cases, an increase in dopant concentration caused higher adsorbed amounts. For F17-11OCB, the amount absorbed was much greater for a fluorinated coating as opposed to a hydrocarbon-coated surface. Also no adsorption of the F3-11OCB could be detected, suggesting that the adsorption process is driven by the oleophobic effect. The results demonstrate that it is possible to manipulate surface properties of liquid crystals using surfactant-like dopants.


Introduction
George Gray's most widely used creation is probably 5CB. [1] A stable, room temperature nematic phase is convenient to work with and so 5CB has become the fruit fly of the liquid crystal world, and nearly every imaginable physical measurement has been done on it. [2][3][4][5] The work reported here is no exception. We have investigated the surfactant-like properties of dopants in liquid crystal phases. However, 5CB has been used instead of water and oleophobic fluorinated groups have been used instead of hydrophobic chains.
This work is partly motivated by curiosity. We want to know whether these nonaqueous systems show adsorbed surface layers analogous to the ubiquitous aqueous surfactant systems and whether the orientational order of the nematic phase influences this adsorption. There are also possible technological applications of this effect. In devices, the interaction of the liquid crystal with surfaces is of particular importance with the surface alignment agents dictating the initial orientation of the liquid crystal. Therefore, surface-adsorbing dopants can influence the material's behaviour even at low concentrations. For instance, while investigating a surface-active polymeric dopant within a liquid crystal host, Bryan-Brown et al. [6] noticed that the dopant induced a weakening of the director anchoring of the bulk nematic phase. The voltage required for switching of a cell between its transmissive and dark states was halved, which in turn would allow faster switching of the nematic orientation or a reduction in power consumption. These would both be significant benefits.
Neutron reflection is well suited to the study of adsorption from solution and there is a large body of work devoted to surface-active materials in aqueous solution. [7] Analogous investigations using nonaqueous or nematic solvents are rare. A previous study of 'F17' (a partly fluorinated version of 11OCB) in 5CB showed strong adsorption at the air interface. [8] The amount of adsorption at the air interface was concentration-and temperature-dependent and was well described by the Brunauer, Emmett and Teller (BET) model, adapted for solutions rather than the gas phase. [9] In the isotropic phase, the adsorbed layer was about 18 Å thick, suggesting that F17 molecules are approximately perpendicular to the surface and less than a complete monolayer. In the nematic phase, the thickness of the adsorbed layer approached 48 Å, the value expected of a bilayer. However, it was not possible to determine whether the phase directly influenced the adsorption because of the sparseness of the temperature-dependent data. The same study found no adsorption at the interface with a CTAB (cetyltrimethylammonium bromide)treated solid substrate.
In this work, we have used recent advances in neutron reflection instrumentation to answer some of the questions raised by the previous study. The FIGARO [10] neutron reflectometer is designed for level, liquid surfaces and benefits from the high intensity of neutrons from the cold source at the Institut Laue-Langevin. It has been used to examine the adsorption of F17 at the isotropic to nematic transition in detail. The lack of interaction with the CTAB surface has been followed by designing a surface to present fluorocarbon rather than hydrocarbon groups to the solution of F17 in 5CB. This has been studied on the INTER reflectometer [11], which uses the second target station at the ISIS Neutron Source and is optimised for interfaces with complex fluids.

Experimental
The 5CB was purchased from Kingston Chemicals (Hull, UK) and used as received. The adsorbate molecules F17 and F3 were both fluorinated versions of 11OCB and their structure and scattering length densities (SLDs) are shown in Table 1. The biphenyl cores of these materials were deuterated to enhance the contrast for neutron experiments. The deuterated hydroxy cyanobiphenyl was synthesised and converted to F17 by Herbert Zimmermann. For the F3, the conversion was done by Germinal Magro. Their purity and level of deuteration (typically 98%) were confirmed by nuclear magnetic resonance. The solutions were made up by agitation in the isotropic phase. The 2% solution of F17 in 5CB was quite close to saturation at room temperature and the possibility of a micellar solution being formed was considered. A small angle neutron scattering measurement (using SANS-2D at ISIS) demonstrated that at least 95% of the F17 existed in molecular rather than micellar form (see Supplemental data).
The neutron reflection measurements were performed with the samples in a temperature-controlled environmental box using a methodology previously described. [8] Although temperatures below 5CB's crystal-to-nematic phase transition were used, supercooling allowed both 5CB and the doped mixtures to remain in the nematic phase at all measurement temperatures below the nematic-to-isotropic phase transition. The samples of F17 in 5CB were contained in a 0.1-mm-deep trough milled out of a silicon block. The bottom of the trough was left unpolished, so it was nonreflective. For the 5CB/air interface, the neutron beam was incident through the air (SLD = 0 Å −2 ) and reflected from the 5CB surface. For the 5CB/solid interface, a polished silicon block was placed on the 5CB surface and the neutron beam was incident through this block. The polished silicon blocks used in the experiment were cleaned using O 2 plasma prior to surface treatment. The two different silane homeotropic alignment agents were the nonfluorinated octadecyltrichlorosilane (OTS), supplied by Sigma-Aldrich (Gillingham, Dorset, UK), and the fluorinated perfluorodecyl dimethylchlorosilane (PFDS), purchased from Fluorochem (Hadfield, Derbyshire, UK). The blocks were submerged in 0.2% and 1% by volume silane in toluene solutions respectively for 15 and 2 minutes before rinsing in toluene to remove excess silane (similar method to refs [12,13]). The blocks were placed in an oven at 80°C for 1 hour to drive off any remaining toluene. After this, the blocks were sonicated in ethanol for 5 minutes, rinsed with water (Millipore, Billerica, MA, USA) and dried with nitrogen gas.
The reflected neutron intensity was measured at two angles and reduced to specular neutron reflectivity by taking the ratio with a direct beam intensity using standard software from the facilities.

Results and discussion
3.1. 5CB/air interface Figure 1a shows the specular reflectivity from 1.9% F17 in 5CB. It has been divided by the reflectivity calculated using the Fresnel formula for the corresponding clean interface (R Fresnel ) to highlight the contribution to the reflectivity of the near-surface structure. The quotient shows the interference features from an adsorbed layer with a neutron SLD that is higher than in both the surrounding bulk phases. The thickness of the 'film' can be estimated from the position of the maximum d~2π/Q max , but it is more accurate to fit the reflectivity calculated for a model SLD profile. [14] The model used for these data is essentially a uniform layer of thickness, d, with the volume fraction of the F17, φ L , with a small roughness applied at the transition between the layers. However, it was found that the fits improved by including a diffuse layer with a low volume fraction of F17, φ E , that decayed exponentially with distance, z, from the interface. Hence, the volume fraction profile of F17 was given by: The SLD of the F17-rich film, ρ F z ð Þ, was determined from its volume fraction: where ρ X is the SLD of the pure solute (F17) and ρ S is that of the solvent (5CB). The mass per unit area or adsorbed amount, q, follows from the density of the pure solute, D X , together with volume fraction of the film integrated over distance: The reflectivity of this model has been fitted to the data and is shown as lines in Figure 1. The roughness between the layer and the bulk phases was fixed at 6 Å throughout. This value gave the best overall fits. The value is slightly larger than the 5 Å expected from the capillary wave model for the known surface tension of pure 5CB. Although an antivibration table was used, it is possible that some residual airborne noise is the cause of the larger value. It was found that best fits were obtained φ L ) φ E and with d Ã e 30 Å. For the 1.9% F17, the volume fractions were typically in the region of φ L e 0:5 and φ E between 0 and 0.04. The fits have been used to determine the uniform layer thickness, d, and the adsorbed amount, q, which are shown in Figure 2.
The thicknesses of the adsorbed layers are slightly less than those found in the previous experiment [8] where 36 Å was found at 25°C for a 1.0% solution. The discrepancy is less at the higher temperatures. The adsorbed amounts are also slightly less than were determined in the previous experiment, the difference being more pronounced at lower temperatures and higher concentrations. These discrepancies suggest that there is a genuine difference in the adsorption between the two experiments. The most likely cause is an error in the concentrations. It was noted that 2.2% is close to saturation at room temperature and it would be difficult to see a white or colourless precipitate in the turbid nematic phase. Hence, it is possible that the F17 had incompletely dissolved or that some precipitation loss might occur after preparation. Figure 2a shows that in the isotropic phase, the thickness of the F17-enriched layer is about 20 Å. This is less than the fully extended length of the F17 molecule (estimated at 24 Å) but substantially greater than its width (5 Å). It suggests that the molecules are roughly perpendicular to the surface, but their orientational order, S, is about 0.5 (S ¼ 3 2 hcos 2 βi À 1 2 where β is the angle between the molecular axis and the director, cosβ e 20=24; S e 0:5). On cooling into the nematic phase, there is not a distinct transition in the thickness, but it steadily increases to much more than the molecular length. The effect is concentration-dependent and is consistent with the formation of a bilayer and possibly multilayers in the nematic phase as previously suggested. [8] The transition is shown more distinctly in the plot of the adsorbed amount, shown in Figure 2b. The amount increases steadily on cooling in the nematic phase. This suggests that there is a larger chemical potential benefit for an F17 molecule to move from bulk phase to surface in the nematic phase, as opposed to in the isotropic phase. This is probably because the homeotropic order at the nematic surface facilitates the fluorinated group sticking out clear of the hydrocarbon solvent (i.e. 5CB). In the isotropic phase, there is a weak maximum at about 40°C that shows beyond the error bars for the 1.9% solution.
Similar features have been reported [15,16] in measurements of surface tension of isotropic 5CB and have been ascribed to nematic wetting of the surface just above the transition. This could also account for the enhanced adsorption around 40°C, since the chemical potential benefit for an F17 molecule to move from isotropic bulk phase to nematic surface would be greater than for an isotropic surface. Figure 3 shows the standard chemical potential scheme and illustrates its effect on the surface adsorption.

5CB/solid interface
The analysis of the reflectivity data from the solid/ 5CB interfaces was done in two steps. First, to characterise the surface layers on a block, the reflectivity from the block in contact with D 2 O and then with pure 5CB was analysed by fitting with a model. For the PFDS-treated silicon, it was found that the natural oxide and the silyl agent, percolated to an extent by 5CB, could be represented by a single layer. Its thickness was 40 Å, and the SLD was 2.4 × 10 −6 Å −2 which is greater than that of the silicon block (2.07 × 10 −6 Å −2 ) and the bulk 5CB (1.43 × 10 −6 Å −2 ) and created a weak fringe in the reflectivity data. For the OTS-treated silicon, the large difference in SLD of the silicon oxide and OTS requires the layers to be modelled separately when characterising the surface. Including the effects of 5CB penetration, their SLDs were 2.81 × 10 −6 Å −2 and −0.28 × 10 −6 Å −2 , and their thicknesses were 28 Å and 32 Å, respectively. These parameters were then used to define the block surface when the reflectivity from the same block in contact with fluorinated dopant (F17 or F3) dissolved in 5CB was modelled.
When the pure 5CB was then replaced with F17 in 5CB, the reflectivity changed significantly, more than expected from the small change in the SLD of the bulk subphase. Beyond the surface coating layers which form a base layer as described above, the dopant distribution was again described by Equation (1). Although in this case, the best fits to the data from F17 in 5CB were obtained with a very thin (~5 Å) uniform layer of high concentration at the surface and a diffuse layer in which the concentration of F17 decayed exponentially over tens of angstrom. This could be interpreted as a layer of F17 molecules in a planar configuration, but this is rather speculative since 5 Å is at the limit of the method's resolution. The fits are shown in Figure 4a, and the corresponding SLDs are shown in Figure 4b. It can be seen that, unlike the air interface, most of the adsorbed F17 is in the diffuse layer. The fits were rather insensitive to the decay length of the exponential, but an estimate of the adsorbed amount can be made from the volume fraction as shown in Table 2.
When the pure 5CB was replaced with 9% F3 in 5CB, the reflectivity changed but only by the amount expected from the addition of the partly deuterated F3 to the 5CB. This suggests that there is no significant adsorption of F3 at the fluorinated silane surface. The OTS-treated block showed a small amount of F17 adsorbed to the OTS. The weak adsorption was modelled ( Figure 5) by a diffuse layer with an exponential decay of the volume fraction of F17, as was used for the air and fluorinated surface on top of a more concentrated adsorbed layer. This suggests that the adsorption is driven by the chemical potential difference between Figure 3. (colour online) Schematic of the standard chemical potential for F17 molecule in bulk phase and at surface of 5CB solvent. The amount of adsorption is also tracked with temperature and will be greater for a greater difference between bulk phase and surface. It can rationalise the greater surface absorption from the nematic relative to the isotropic phase and the weak maximum in the isotropic when the surface is still nematic but the bulk phase is isotropic. an F17 molecule at the OTS surface and in the 5CB subphase. However, the difference is less than that of the PFDS surface.
The determination of the adsorbed amount for F3 has a significant uncertainty associated with it from the similarity in the SLD of F3 and PFDS. However, the data could not be modelled with an adsorbed amount greater than (0.1 mg m −2 ) and are best fitted with zero. This can be interpreted as a consequence of the much lower number of fluorine atoms on F3 than on F17, which causes the difference in standard chemical potential between the molecule staying in the bulk phase and near the surface to be far smaller than for F17. The lack of adsorption of F3 at the fluorinated surface and the weak adsorption of F17 at the nonfluorinated OTS surface both support the conclusion that the oleophobic effect, which derives from the incompatibility of fluorocarbon and hydrocarbon chains, drives the adsorption of these fluorinated dopants.
There is a distinct difference between the profile of the F17 adsorbed from 5CB at the air/5CB interface and at the solid surface as shown in Figure 6. At the air/5CB interface, the F17 is in a well-defined layer with a thickness of one molecular length in the isotropic phase and approaches a bilayer thickness in the nematic phase but the diffuse layer has a very low Note: # The volume fraction of the dense uniform layer was parameterised as the maximum value of 1; however, the roughnesses applied to the layers (given in the Supplemental data) reduced the maximum as shown in Figure 6.  (possibly zero) concentration. In contrast, at the fluorinated solid, most of the F17 is in a diffuse layer. In the framework of the BET model for the adsorption of a solute at the surface of a solvent, the amount adsorbed in the first layer is defined by an equilibrium constant K S , and it also defines the concentration of the solute at the maximum of the diffuse layer. With some solute adsorbed in the first layer, the diffuse layer arises from a weak preference for solutesolute contact rather than solute-solvent contact. The decay length, , of the solute concentration in the diffuse layer is defined by the monolayer thickness, d m , and the solute on solute equilibrium constant, K L .
where c is the concentration of the solute in the continuous phase. This model is consistent with the results for the solid surface where the fluorinated surface has a higher K S than the OTS surface, so the starting point of the decay is higher but the decay length is of similar magnitude. However, it is not consistent with the results from the air interface, where, despite the strong adsorption in the first layer, the diffuse layer is almost absent. This suggests that the close packing and strong homeotropic alignment at the air/nematic interface cause some deviation from the BET model. At high concentrations in the nematic phase, the oleophobic effect [17,18] drives the fluorinated groups to pack densely at the air surface as shown in the cartoon in Figure 7. The second layer is able to adsorb onto it quite easily because the electric dipoles on the cyano groups (−CN) are able to overlap with those in the first layer. However, this does not work for the third layer which is presented with a fluorocarbon-rich layer with the negative ends of the dipole pointing to the bulk phase. Since both ends of the F17 molecule have a negative charge due to the electronegative elements (N and F), they are not attracted to the bilayer so the concentration in the diffuse layer remains low. This effect does not occur for the solid interface studied here because the amount of F17 adsorbed in the first stratum is much lower, and so there is space for the molecules in each subsequent stratum to find a favourable position with the dipoles overlapping rather than head-to-head.

Conclusions
The adsorption of a partially fluorinated 11OCB molecule, F17, from solution in 5CB has been shown to occur at both the air surface and a solid surface coated with a fluorinated silane. The adsorption at a solid surface coated with a nonfluorinated silane, OTS, was very much weaker. For a similar molecule with a lower degree of fluorination, F3, there is no measurable adsorption at a fluorinated silane-treated surface.
The temperature dependence of the adsorption has been studied, and a distinct change in the adsorption behaviour has been demonstrated at the isotropic-to-nematic phase transition. The significant increase in adsorbed amount upon cooling within the nematic phase indicates the enhancement of adsorption at the surface of the orientationally ordered phase. A qualitative scheme has been proposed for the standard chemical potential of F17 in 5CB, both in the bulk phase and at the surface. Changes in chemical potential can rationalise the temperature dependence of adsorption in the nematic phase and the weak adsorption maximum in the isotropic phase at around 40°C where nematic wetting of the surface is believed to occur.
While the F17 forms a layer with a high concentration at the air/5CB interface, with adsorbed amounts in the range of 1.5-3.1 mg m −2 , the diffuse layer formed at the fluorinated solid surface overall contains a larger amount of material, up to 4.9 mg m-−2 . The adsorption at the solid surface also extends further into the bulk 5CB with its exponential decay length fitted by a model as 75 Å. The adsorption at the solid surface can be understood in the context of the BET multilayer model, whereas at the air interface the highly concentrated first-layer adsorption appears to facilitate a concentrated second layer but strongly hinders subsequent layers from adsorption. It remains to be seen whether this type of fluorinated liquid crystal dopant can be used in a device to form a slippery surface and improve the efficiency of switching states.
5CB remains at the start of many developments in thermotropic liquid crystal research, which is testimony to George Gray's invention. We plan to continue to explore methods to modify liquid crystal/ surface interactions using soluble dopants in 5CB. Figure 7. (colour online) Suggested arrangement of the F17 molecules amongst the 5CB in a bilayer at the air/nematic interface. The dipoles from the fluorinated groups point into the bulk phase and oppose adsorption of subsequent layers. F17 molecules are depicted with the expected bend at the oxygen link projected into the plane of the paper, and the 5CB solvent is omitted for simplicity.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
LM and WdV gratefully acknowledge the funding provided by EPSRC under grant number EP/H0148611.

Supplemental data
Supplemental data for this article can be accessed here.