Thermotropic liquid crystals (LCs) are remarkable materials. The discovery of substances in which the optical properties of birefringent crystals are combined with fluid-like behavior have enabled our modern world of information display [1
]. Display applications have taken advantage of LCs, beginning with the first twisted nematic display in 1972. Today, liquid crystal displays (LCDs) are ubiquitous in most electronics, from smartphones to large-size television screens. The ubiquity of LC displays arises in part from their responsiveness to external stimuli allied to chemical stability. These valuable LC properties have led to increasing research efforts [2
] into other applications, including chemical and biological sensors.
The unique combination of LC responsiveness to the environment and the striking optical effects that allow the rapid visualization of this response facilitates the use of LCs in sensing applications. The LC responds to several different classes of molecules, including surface-active agents such as lipids and surfactants [5
] and non-surface-active molecules such as gas vapors [10
] and can be tailored to respond only to specific antigens [9
Several recent books and reviews have summarized and categorized the large variety of approaches to biosensing. Some of these focus on electrochemical, bioelectronic, piezoelectric, cellular and molecular biodetection approaches [18
] and some are devoted specifically to optical biosensors [24
]. Most approaches fail to meet the full range of required or desirable sensor characteristics. Recently, attention has been directed to liquid crystal-assisted biological and chemical sensors [2
]. As first proposed by Abbott et al. [15
], such advanced functional materials hold great promise of overcoming many challenges because LCs readily respond to a variety of chemical and biological agents by reorienting their constituent molecules, which can be easily determined by a beam of light passing through the LC slab during a sensing event. An example of such a transition is shown in Figure 1
Although we focus on sensor platforms that are based on thermotropic LC materials in this review article, here we also briefly mention another sensing platform that is based on lyotropic LCs. The variety of molecules that are capable of forming lyotropic phases is extremely rich and complex mixtures can be formulated for biomedical applications, drug delivery vehicles and biosensors [33
]. Water-soluble discotic shaped molecules may form lyotropic chromonic LC (LCLC) phases [34
]. LCLC find applications in biosensing, where larger agglomerates of antibodies, viruses or even cells may need to be identified [35
The principle behind this type of sensing is based on visualizing a director defect which is produced when an object in the medium is larger than the extrapolation length , where K is the director distortion elastic constant in the order of 1–10 pN and W is the anchoring strength typically found in the orders of 10−3–10−5 J·m−2. If the size of the microbe before binding is smaller than b but becomes larger after clumping with other microbes via binding antibodies, a defect appears that can be detected optically.
Lyotropic-based biosensors were commercialized by Crystal Diagnostics Ltd. (Broomfield, CO, USA) [37
] and are currently accredited for the rapid detection of the pathogenic strain of Escherichia coli (E. coli)
O157 and salmonella in food and for screening of Listeria monocytogenes
on environmental surfaces.
Progress was made in improvements of the homogeneous and stable alignment on anisotropic surfaces [38
]. This is a critical requirement for the applications mentioned above to mitigate the occurrence of false optical signal. Berride et al. reported that addition of small chiral organic molecules—such as amino acids—to achiral disodium chromo glycate (DSCG) phases induces a chiral phase through the formation of tactoids at the interface between isotropic and nematic phases [39
]. This interesting effect on the LC director configuration can be used in enantio-selective detection or chirality discrimination.
An interesting lyotropic LC class is realized by the self-assembly of short DNA oligomers [40
]. It would be also interesting to design a lyotropic LC composed of aptamers and compare the changes in LC textures after introducing antigens that bind to these molecules, with any changes when adding antigens that do not. Such an approach could lead to the development of novel lyotropic aptamer biosensors.
Thus, lyotropic LCs are a very interesting sensor platform with one major advantage: they can be biocompatible, so cells and biological molecules may float within the LC without being destroyed. However, comparatively few studies focus on lyotropic LCs, so the rest of this review is devoted to thermotropic LC sensors.
2. Sensor Formats
In most thermotropic LC sensors, the analytes remain outside the LC and instead adsorb on its surface. The molecules that make LCs are very sensitive to surfaces they come in contact with due to the delicate balance between the energies of bulk elastic deformations and surface anchoring of the director [42
]. Abbott et al. applied this sensitivity to use LC materials as a transducer element of chemical and biological sensors [7
]. LC materials modulate the light that propagates through it due to long-range order in the arrangement of optically anisotropic molecules and, additionally, the LC acts as an amplifier of the interfacial changes; thus, an LC-based sensor provides an optical output signal as schematically shown in Figure 2
To date, two major thermotropic LC-based sensor formats have been utilized in research experiments: thin LC films in contact with aqueous samples and aqueous LC emulsions. Other designs have been recently proposed, including LCs in micro-capillary tubes [43
] and fibers [45
Many different methods have been proposed to increase sensor sensitivity, including adjusting the physical properties of the LC and the phases it displays. The nematic phase, with orientational and no positional long-range order, is the most common and, because it usually flows readily, often shows the fastest response. Recently, other LC phases have started attracting interest for specialized modes of sensing. These sensing LC states include, but are not limited to, smectics [17
], cholesterics [48
] and blue phases [51
In this review, we will discuss these different sensing modes. Further, we will discuss some of the limits on analyte properties, along with the influence of solvent and co-solutes. The adaptation of this technique to specific sensing of analytes is critical for its utility; several examples of how this has been done will be given. Finally, the role of numerical simulations in understanding the sensing mechanisms at a molecular level will be highlighted.
2.1. Freely-Suspended LC Films
In the first LC sensor proposed by Abbott et al., LC films were supported by a transmission electron microscopy (TEM) grid and a solid glass substrate coated with an alignment layer [15
]. Usually this alignment layer is designed to be homeotropic, aligning the director perpendicular to the glass surface. Later, Hartono et al. proposed freely-suspended LC films supported only by a TEM grid [53
]. Since both these types of film fill the cells of TEM grids, their thicknesses are approximately equal to the thickness of the grid, which is typically 20 µm (see Figure 3
). Both approaches have advantages and disadvantages.
Laboratory preparation of freely-suspended LC films is much less time-consuming than the preparation of the additional glass substrates with alignment layer. As shown in Figure 3
, before binding of surfactants to the LC/aqueous solution interface, the alignment of the nematic LC film at that interface is planar and the alignment at the LC/air interface is homeotropic. These initial boundary conditions are well-reproducible across experiments and in widely varying external conditions, such as throughout the whole range of temperatures within the nematic phase of the LC film and for any air humidity levels.
Freely-suspended films can be submerged entirely in an aqueous solution but this process must be performed carefully. Since the thermotropic LCs are oily hydrophobic substances, just before submerging a meniscus is formed between the water and the TEM grid that contains the LC films and the water often flushes over the grid as soon as the grid is pushed deep enough. This flushing may displace the LC in the grid causing some of the cells to be overfilled and others to be under filled. This undesired effect may be minimized by submerging the TEM grid in cold water, which upon the initial contact makes LC films stiff: for example, 4-cyano-4′-pentylbiphenyl (5CB) may become a solid or 4-cyano-4′-octylbiphenyl (8CB) may transition to smectic-A. After the water is warmed to room temperature (for experiments using e.g. 5CB as in an example shown in Figure 4
f–j or to mammalian body temperature of ~37 °C (for experiments using e.g. 8CB as in an example shown in Figure 4
a–e, the LC films return to the nematic phase. Typically, the separate films in cells of the TEM grid appear more uniform in color, indicating better thickness uniformity.
As reported by Popov et al. [55
], freely-suspended LC films are not suitable for experiments where the grids are fully immersed in aqueous solutions containing certain surfactants. For example, the addition of the non-ionic oil-soluble emulsifying surfactant Triton X-100 into water did not facilitate the planar-to-homeotropic transition but instead led to 5CB film rupture. One must also be careful with freely-suspended cholesteric LC films, as they may form LC microlenses instead of remaining flat when submerged under water [56
2.2. LC Films with Solid Supports
After the first phase of “trial and error” experiments using freely-suspended LC films, it may be desirable to introduce the solid glass substrate to better support one of the LC surfaces. The solid substrate is typically coated to achieve homeotropic LC alignment, as provided by air but glass substrates can be alternatively treated to provide planar alignment. In some cases, the solid substrate provides not only the alignment layer but are also functionalized with sensing components designed to bind with target analyte, after it diffuses through the bulk of the amplifying LC film and change the director configuration [57
2.3. Grids to Hold LC Films
TEM grids are commonly used for supporting LC films designed for sensing chemical and biological molecules. They are convenient and readily available commercially but they are also limited by their intended purpose in electron microscopy. TEM grid cells are usually square or hexagonal in shape and approximately 20 µm deep. TEM grids are usually made of gold, nickel or copper, which restricts the anchoring energy of LCs with the grid walls. Additionally, copper grids tend to oxidize and degrade quickly when in contact with water. Thus, it may be desirable to prepare custom grids designed specifically to hold the LC in LC-based sensors. For example, Bedolla et al. [58
] fabricated chemically patterned micro-wells of precise depths (0.7–30 µm) to hold strained nematic LC films for the sensing of low concentrations of toluene vapors.
2.4. LC shells and Droplets
So far, sensing applications have mainly focused on liquid crystals in flat films, largely due to the convenience of fabrication and visualization of the liquid crystal textures through microscopy. Recent research has explored the use of liquid crystals in curved geometries, such as droplets and shells [59
]. Monodisperse LC droplets and shells are often achieved with microfluidics chips [46
]. For sensing applications, they can be used as produced or can be functionalized. Droplets and shells have the advantages of using small amount of fluid, while having a high throughput of consistent product that can be easily tuned just by the adjustment of flow rates. The chips can either be in the form of simple poly(dimethylsiloxane) (PDMS) rubber, normally fabricated through soft lithography processes, or glass capillary-based chips made by techniques developed originally by the Weitz group [62
Depending on the geometry used, microfluidic chips can be used either for simple droplet production [63
], for the generation of filaments [64
], or for the creation of liquid crystal shells [46
]. The interface between a liquid crystal shell or droplet and the surrounding medium affects the LC alignment and thus leads to sensing the presence of surface-active molecules, much in the same way as for thin films. For example, Humar et al. [65
] used liquid crystal droplets to visualize the adsorption of sodium dodecyl sulfate (SDS) to a 5CB-water interface by observing the anchoring transition from planar to homeotropic/radial. Similarly, Noh et al. [61
] visualized the orientation of stabilized 5CB shells as a function of surfactant concentration.
2.5. LC in Capillaries and Fibers
In addition to flat and spherical interfaces, cylindrically-shaped liquid crystals, created either by filling capillary tubes or in polymer fibers with LCs, can be also used in sensing platforms.
Due to the capacity of glass to be functionalized to create either planar or homeotropic alignments, a glass capillary can easily be customized to consistently create a desired initial texture before the sensing event. Capillary-based platforms have been used for visualizing the presence of biomolecules. Kim et al. initially presented work on a sensor consisting of a liquid crystal confined in a capillary [43
], treated to produce homeotropic radial anchoring, to which surfactant molecules, such as 1,2-dioleoyl-sn
-glycero-3-phosphoglycerol (DOPG), were adsorbed from a confined solution in the capillary and between LC regions, as shown in Figure 5
A. The sensor was able to detect the presence of biomolecules such as trypsin, poly-l
-lysine and phospholipids.
Furthermore, they demonstrated detection of bile acids, such as cholic acid [44
]. The presence of the target biomolecule disrupts the homeotropic alignment of the LC caused by alkyl trimethylammonium bromides, such as CTAB, or SDS, displacing the surfactant and restoring a planar texture at the aqueous-LC interface as schematically shown in Figure 6
Another emerging use of liquid crystals in sensing applications is by encasing the liquid crystal in polymer fibers. The advantages to this approach are clear, as the liquid crystal is stably and securely contained within the polymer, thus creating a portable, robust, easy to use platform.
One of the first methods for fiber production was through a phase separation process, where the polymer and liquid crystal would be mixed together into a single solution and, as the mixture is spun or sprayed, the liquid crystal would separate from the polymer, remaining in the matrix while still forming distinct liquid crystal droplets [66
]. A one-step process that is relatively easy to perform has allowed the creation of liquid crystal-based textiles for purposes such as temperature sensing, a potentially useful tool in medical diagnostics [67
]. Alternatively, it has become increasingly popular to produce coaxial fibers, with a liquid crystal core surrounded by a polymer sheath. Extremely thin coaxial fibers can be rapidly and consistently formed through electrospinning techniques, enabling a mobile sensing platform that also produce optical responses visible to the naked eye [11
]. Kim et al. [71
] and Reyes et al. [11
] demonstrated the use of liquid crystals encapsulated in a permeable polymer fiber, such as poly(vinylpyrrolidone), for the sensing of volatile organic compounds (VOCs), as shown in Figure 7
The response to VOCs was visible even to the naked eye, showing a change from scattering to transparent in the presence of the organic vapor. It was hypothesized that the change in scattering occurs as a result of the mobility of the VOCs through the polymer sheath due to the porosity of the polymer, thus lowering the clearing point of the liquid crystal contained within. Complete transparency is achieved only above 3% of VOC, when the solvent vapor condenses in the fibers and fills the interstitial area with the fluid solvent, thus decreasing the refractive index variation. Removal of the VOCs leads to the system reverting to its initial state over time.
2.6. LC-Assisted Direct Visualization of Graphene Features
Graphene is a transparent and flexible conductor that holds promise for various applications, such as in solar cells, batteries, catalysis, hydrolysis, biosensors and bio-imaging [73
]. Particular interesting applications of LC sensing films are in detecting orientations of graphene and of other similar two-dimensional (2D) sheets and their boundaries [75
]; visualizing flake deformations [78
] and identifying their deformation-induced chirality [79
Kim et al. proposed a simple method for the visualization of arbitrarily large graphene domains by imaging the birefringence of a nematic liquid crystal film that covers the graphene layer [75
]. This method relies on a correspondence between the orientation of the liquid crystals and that of the underlying graphene (see Figure 8
). Caused by strong π-stacking interaction, the hexagons present in the graphene film might induce highly ordered packing of liquid crystals containing benzene rings and alkyl chains.
Shehzad et al. [77
] reported the visualization of grains and boundaries of chemical vapor deposition-grown molybdenum diselenide and tungsten diselenide on silicon using optical birefringence of nematic LC films that cover 2D layers.
Basu et al. [79
] have demonstrated that graphene nano-flakes may serve as chiral dopants for liquid crystals due to strain and edge chirality of the flakes. Thus, nematics and smectics may be used for the detection of graphene deformations and the type of flake boundaries (see Figure 9
). The authors also found that LC doped with graphene may have enhanced dielectric anisotropy, faster electro-optic response and enhanced spontaneous polarization [80
6. Computational Approaches to LC Sensor Designs
In recent years, huge progress has been made in the development of novel advanced functional nanomaterials. Many successful experiments have been performed in search of functional sensor designs with predetermined schemes for a reliable and informative response. Typically, the preliminary design ideas are quite simple but very often unexpected behaviors result, especially when the analytes are complex biological molecules. Clearly, optimizing LC sensing devices requires tracking chemical reactions and physical interactions on an atomic level. Thus, systematic computational chemistry and molecular dynamics methods may prove to be very useful approaches.
6.1. Computational Chemistry Methods
Optimization of a particular LC sensor platform is often a rather challenging task due to an involvement of a high number of experimental parameters, some of them not even identified in initial experimental trials. To address this challenge, Szilvási et al. [124
] proposed an approach of integrating cycles of computational chemistry, organic synthesis and physical property evaluation for the efficient design of novel chemo-responsive LCs as schematically represented in Figure 23
In this work, Szilvási et al. used electronic structure calculations of the binding of nitrile-containing mesogens to perchlorate salts. This approach predicted that selective fluorination can reduce the strength of binding of nitrile-containing nematic LCs to metal-salt decorated surfaces and thus generate a faster reordering of the LC in response to competitive binding of dimethyl methylphosphonate (DMMP) gas. The authors subsequently synthesized and tested several fluorinated compounds and confirmed their theoretical predictions of their “coordinately saturated anion model” (CSAM). In this model, the perchlorate salt ion is explicitly described since, as shown in [125
], the choice of the salt anion can influence the interaction of nitrile-containing mesogens with metal-salt decorated surfaces. The CSAM model showed that selective fluorination of the 5CB mesogens reduces the binding energy of the nitrile group to metal cations and thus leads to a substantial response to DMMP.
6.2. Molecular Dynamics
In recent years, huge progress has been made in the development of novel advanced functional nanomaterials. Many successful experiments have been performed in search of functional sensor designs with predetermined schemes for a reliable and informative response. Typically, the preliminary design ideas are quite simple but very often unexpected behaviors result, especially when the analytes are complex biological molecules. Clearly, optimizing LC sensing devices requires tracking chemical reactions and physical interactions on an atomic level and thus systematic computational chemistry and molecular dynamics methods may become very useful tools.
Atomistic Molecular Dynamics (MD) calculations often require huge computer resources, even in the case of optimized and parallelized simulated systems, which typically need to contain a very large number of interacting molecules to properly describe realistic physical environments and events. One way of simulating large biological systems is to depart from the atomistic level of details and simplify the structure of molecules by, for example, representing phospholipids, proteins and other relevant molecules in coarse-grained (CG) simplification [126
] and reducing the rod-like LC molecules to spheroids [129
Luckily, the famous Moore’s “law,” which predicts the computational power of computers doubles approximately every 18 months, is still in effect. Thus, simulations are expected to be effective for increasingly more complex systems. Larger number of atoms and longer simulation time frames become accessible to reproduce realistic binding/unbinding events that may occur in biosensor systems [120
]. Recently, complete fully atomistic models of capsids such as of HIV-1 and Hepatitis-B viruses were developed [134
], which may facilitate much more accurate modeling of interaction events at biosensor/capsid interfaces. MD simulations were also employed in revealing the details of the organization of LC molecules in thin films at interfaces [137
] (See Figure 24
) and were used in predicting their stability and rupture mechanisms [140
]. When simulations results agree well with real experimental observations, the MD serves as a “computational microscope” [141
] that allows visualization of objects that often no other tool can.
The realism of MD simulations is highly dependent on the accuracy of the chosen force field (FF). Many FFs were developed in the last decades but perhaps the most applicable to modeling of LC biosensors is the FF named Optimized Potentials for Liquid Systems (OPLS). Incremental updates to this FF allowed in recent years the more accurate reproduction of phenomena occurring at liquid interfaces [142
]. An improved version OPLS3 is now available. As reported by Schrödinger, LLC (New York, NY, USA) [143
], this new version adequately covers the medicinal-chemical space by employing over an order of magnitude more reference data and associated parameter types relative to other commonly-used small molecule force fields (e.g. MMFF, OPLS_2005, etc.). Electronic coarse-graining is another important approach for more realistic reproduction of biologically-relevant forces such as many-body polarization and dispersions [144
LC biosensors were reported to identify enzymatic reactions [145
]. Such chemical reactions cannot be simulated with conventional FFs. Thus, recently a new reactive force field (ReaxFF) was developed [147
]. Improvements to QM/MM (quantum mechanics/molecular mechanics) methods in multi-scale modeling were reported as well [149
]. Development of such emerging MD tools [150
] is crucial for accessing the full range of possible biochemical events for the optimal design of sensors, drug-delivery systems, immunoresponse manipulation and so on.
7. Summary and Future Directions
The field of chemical and biological sensing using thermotropic LCs has made great progress during the 20 years since the introduction, by Gupta et al. in 1998 [15
], of the method of optical amplification of ligand-receptor binding using LCs. Many sensing platform modifications and improvements have been suggested since then. Many milestones have been reached and reported in numerous articles [7
Research achievements strongly suggest that the LC approach is indeed a very promising one but further challenges in the development of functional chemical and biological sensors that will be competitive with solutions already on the market need to be addressed. These include:
Sensing LC interfaces need to be robust. This is especially important for platforms with flat sensing surfaces, in which LC films need to be protected from rupture or becoming washed away into the aqueous solution [17
Detection limits need to be as low as or lower than the sensitivities of existing non-LC sensors. Additionally, a linear sensor response over the interesting range is desirable [6
New types of detection modes that are not based on measurements performed at often slowly-achieved equilibrium states, may become very useful. Further, the final equilibrium states induced by different analytes may appear similar and even indistinguishable but the dynamics in achieving that final equilibrium state may reveal the presence of a specific analyte. Thus, non-equilibrium processes deserve serious attention in the future development of LC-assisted sensor platforms [5
One of the most important of customer requirements is that the sensors need to be extremely specific towards a single kind of analyte of interest. Research that demonstrates true and reliable specificity of LC-assisted sensors employing antibodies, aptamers and similar molecules remains scarce [14
]. Reproducing and simplifying the biomimetic mechanisms of molecular recognition may serve as an efficient way of ensuring the required high specificity of biosensors [36
Another important market requirement is that the final sensor platform must be compact and especially easy and ready to use without much preparation. Thus, prefabricated devices with long shelf-lives are in high demand. At least two possible directions have been suggested: Ionic LCs may prevent fragile enzymes and antibodies from denaturing [99
]. Aptamers, much more stable than antibodies, may provide the required level of specificity [121
Fundamental research on the behavior of LC-based sensors towards specific analytes in the presence of interfering species [55
] is only at the beginning. This extremely important problem will need to be extensively addressed prior to any serious attempts at commercializing new types of sensors. It is likely that LC/analyte interactions, as well as chemical reactions, will need to be clearly understood and visualized at the molecular level for the successful development of superior future sensors [124
In this review, we have summarized the most recent progress on the development of novel LC-based sensors for chemical and biological analytes. Even though this field is still in its infancy, sensor technologies revolving around LC materials are being patented and commercialized [37
]. On a broader scale, we recognize that novel platforms (not necessarily LC-based) for specific, reliable, inexpensive and rapid sensing are in growing demand [151
]. These needs are validated by various companies such as ProXentia S.r.l. (Milan, Lombardy, Italy) [152
] and Dynamic Biosensors Inc. (Martinsried, Munich, Bavaria, Germany) [153
] and many others [154
], entering this market. Due to the dramatic responsiveness of LCs to various stimuli [155
], LC-based solutions hold the promise for significant breakthroughs in sensor applications. For biological applications, LC materials deserve additional attention due to the omnipresence of LC phases in various developing tissues of the brain, liver, kidneys and other organs [156
]. All these unique and exciting features of LCs provide a myriad of opportunities for the development of novel smart functional materials and place these self-organizing substances at the forefront of ongoing global research.