betaFeedback

WorksPhysics and Astronomy

On Modeling Tear Breakup Dynamics with a Nematic Lipid Layer

Abstract

One of the main roles of the lipid layer (LL) of the tear film (TF) is to help prevent evaporation of the aqueous layer (AL). The LL thickness, composition, and structure all contribute to its barrier function. It is believed that the lipid layer is primarily nonpolar with a layer of polar lipids at the LL/AL interface. There is evidence that the nonpolar region of the LL may have liquid crystalline characteristics. We investigate the structure and function of the LL via a model of the tear film with two layers, using extensional flow of a nematic liquid crystal for the LL and shear-dominated flow of a Newtonian AL. Evaporation is taken into account, and is affected by the LL thickness, internal arrangement of its rod-like molecules, and external conditions. We conduct a detailed parameter study with a focus on the evaporative resistance parameter, the Marangoni number, and primary liquid crystal parameters including the Leslie viscosities and director angle. This new model responds similarly to previous Newtonian models in some respects; however, incorporating internal structure via the orientation of the liquid crystal molecules affects both evaporation and flow. As a result, we see new effects on TF dynamics and breakup.

Introduction

With each eye blink, a thin liquid tear film is established that helps protect and nourish the ocular surface, and helps provide a good smooth surface for clear sight. The tear film (TF) is a multi-layer film, and each layer plays a role in TF dynamics [1]. Despite the importance of the TF, there is still much that is not understood. This paper develops a new model for how the TF may fail, which is often called tear breakup (TBU). The purpose is to better understand the function and structure of the outermost layer of the TF, called the lipid layer (LL). A prime motivation for understanding TF dynamics and TBU is that these phenomena are thought to contribute to the initiation and progression of dry eye disease (DED), which affects millions of people [2, 3]. For comprehensive views of clinical aspects of DED and the tear film see Refs. [4, 5, 6, 7, 8]. Reviews related to the tear film include Refs. [1, 9, 10]. In this introduction, we now describe the overall structure of the LL. We then introduce previous work on properties, proceed to mathematical models of the tear film and its dynamics, and finally discuss details of the LL and our goals for this paper.

Overall structure of the tear film

At the anterior interface with air is an oily LL that retards evaporation [11, 12], and helps to retain a smooth, well-functioning tear film [13]. The LL also contains polar lipids that are surface active and lower the surface tension between the LL and the aqueous layer (AL), which is posterior to the LL [14, 15]. The AL consists mostly of water [16], but it also contains various ions and proteins [11]. At the corneal surface, which is posterior to the AL, there is a region with transmembrane mucins protruding from the epithelial cells of the cornea or conjunctiva. This forest of glycosolated mucins is called the glycocalyx. It is generally agreed that the presence of the healthy hydrophilic glycocalyx prevents the tear film from dewetting the ocular surface [17, 18, 19].

The overall thickness of the TF is a few microns [20], while the average thickness of the LL is on the order of tens to 100 nanometers when the eye is open [10, 13, 21, 22, 23]; the thickness of the glycocalyx is a few tenths of a micron [19, 24]. The overall structure is reformed on the order of a second after each blink in a healthy tear film [10, 14, 25, 26].

TBU occurs when a thinned region forms in the TF. Clinically, this is defined as the first dark area that is observed in the fluorescent TF following instillation of fluorescein dye [27]. Various mechanisms are thought to cause different types of TBU, or to combine in various ways: evaporation of water from the AL [7, 28, 29], Marangoni-driven tangential flow (from non-uniform polar lipid surface concentrations) [30], or dewetting-driven tangential flow (from uneven ocular surface conditions) [31, 32, 33]. Here, we use the phrase tangential flow to mean flow along the eye. The evaporation of water from the AL causes relatively slow thinning [34], while one may expect rapid thinning to be explained by Marangoni-driven tangential flow [30] or dewetting-driven tangential flow. Concentration gradients from surface active polar lipids in the LL induce shear stress at the LL/AL interface and drive outward tangential flow via the Marangoni effect [14, 15]. Dewetting from irregularities in a corneal surface region has been hypothesized to generate outward tangential flow from internal pressure gradients inside the AL due to van der Waals type forces [31, 35, 36]. A related term is full-thickness tear film breakup (full thickness-TBU), which is when the AL has thinned to the point where the LL and glycocalyx touch [29, 37]. It is our belief that the strongest evidence to date regarding mechanisms for TBU are evaporative and Marangoni-driven thinning and their interaction [38, 39]; unambiguous evidence of van der Waals-driven dewetting is not yet in hand in our opinion [29]. We now discuss studies of these mechanisms in more detail.

The aqueous part of tear fluid is primarily supplied from the lacrimal gland near the temporal canthus (corner) and the excess is drained through the puncta near the nasal canthus [40]. Water lost from the tear film due to evaporation into air is an important process [11, 41, 42]. We believe that this is the primary mechanism by which the osmolarity (essentially, the total concentration of ions on a molar basis) is increased in the tear film [10]. Some water is supplied from the ocular epithelia via osmosis when the osmolarity increases above the isotonic level (the so-called hyperosmolar state) [10, 41, 43, 44, 45]. According to the Dry Eye WorkShop (DEWS) reports, the osmolarity is important because it plays an important etiological role in the development of dry eye disease [28]. An accepted view in the DEWS reports is that repeated exposure to sufficient hyperosmolarity in tears, as may occur in TBU, can result in inflammation and ocular surface damage in a vicious cycle that may lead to DED. There is still considerable ongoing research to more fully understand the causes and etiology of DED; some mathematical modeling results discussed in this paper help to quantify the relationship between tear breakup and hyperosmolarity.

Measurements and properties

Both TF instability and hyperosmolarity are important to study because they are proposed as etiological causes of dry eye disease [6, 7]. An osmolarity difference between the corneal epithelium and AL induces osmotic flow from the cornea to the TF [10, 46]. TF osmolarity may be measured in the inferior meniscus clinically [47]; the healthy range is 296-302 mOsM [47, 48, 49]. Dry eye measurements in the same location can reach 316-360 mOsM [48, 50, 51, 52] but estimates for the TF over the cornea reach 900 mOsM or higher [10, 46, 53, 54]. High levels of TF osmolarity are associated with pain, inflammation and cellular changes [53, 55, 56]. In support of these potentially high levels of TF osmolarity over the cornea, mathematical models without spatial variation have estimated peak osmolarities up to ten times the isotonic concentration [10, 43]. The modeling work of Peng et al. [46] found that evaporation elevates osmolarity in breakup regions.

TF thinning rates have been measured experimentally or estimated in many studies. A few experimental methods include spectral interferometry [34, 42, 57], an open chamber [58], an evaporimeter [59], and averaging pure water and functioning LL rates over the cornea obtained by heat transfer analysis and thermal imaging [60]. Parameter estimation schemes were developed in Luke et al. [39, 54] for fitting partial differential equation (PDE) models to experimental fluorescent intensity distributions [61]. They found evaporative thinning rates ranging from 36.9 to −4.91 μ\mum/min (negative indicates thickening) and overall TF thinning rates ranging from 23.5 to 1.85 μ\mum/min. These estimates were quite similar to previous experimental spot-thinning rates that did not specifically seek TBU instances [57].

Tiffany and Dart [62] measured values of the viscosity of human meibum (the material expressed from the meibomian glands inside the lids) using a capillary tube and they estimated values from 9.7 to 19.5 Pa s at 308 K. This is a representative value for the temperature on the surface of the tear film [63]. In a series of papers by Tiffany and coworkers, the human tear film has been shown to have non-Newtonian properties [64, 65, 66]. Tiffany [66] found that whole tears, that is, tear sampled directly from the eye, are shear thinning. The shear thinning has been fit with empirical models, namely Cross [66] and Ellis models [67]. Tiffany [64] also found weak elastic effects; Tiffany and Nagyová [68] found that removing all lipids from tears caused tear fluid to become Newtonian. The rheology of the LL depends on temperature [69]; in vitro observations showed temperature dependence of meibum on saline for monolayers [70, 71] and for bulk samples [72]. The elastic contributions to viscous response was small above 305 K for these measurements [70]. King-Smith et al. [23] used high-resolution microscopy to observe LL dynamics, and they saw fine structure of the LL during the interblink period that included islands of lipids.

The surface tension of the tear/air interface was measured to be 46.2 mN/m by Miller [73] in vivo using a special contact lens; this is about 2/3 that of an air/pure-water interface. Pandit et al. [65] measured the surface tension of whole tears with a capillary tube and obtained 43 mN/m for unstimulated tears and 46 mN/m for stimulated tears; Miller [73] likely measured stimulated tears. A value of 45 mN/m is often used in theoretical studies. The components responsible for lowering the surface tension from that of pure water has been the subject of a number of papers [68, 74, 75, 76]. A challenge for this measurement is the dependence of surface tension on surface active components like phospholipids and other lipids [77, 78, 79] and still other species [80]. Detailed equations of state for studying the Marangoni effect from these species is not available for the tear film in vivo to our knowledge, and this is an ongoing area of research.

Mathematical modeling

A variety of mathematical models have incorporated various important effects of tear film dynamics as has been recently reviewed [10, 43, 81]. The most common assumptions for these models are a Newtonian tear fluid and a flat cornea [14, 82]. One category of models treats the flow over an open eye-shaped domain; these models have used lubrication theory with two space dimensions and time as independent variables [83, 84, 85, 86, 87]. These papers were on a stationary domain, and emphasized the effects of supply and drainage, osmolarity in various locations in the tear film, and the overall appearance of fluorescein imaging of the tear film. Some initial studies have examined flow on a blinking domain [81, 88, 89], but they have yet to incorporate the same level of effects as in the static boundary papers. There have also been useful compartment models developed for tear film dynamics [44, 90]. These models use compartments of the tear film as dependent variables in systems of ordinary differential equations (ODE) and thus avoid the computational difficulties of solving PDEs on 2D domains. The models can be solved over many blinks and give insight into the osmolarity and other variables at various operating conditions.

There are quite a few models that have been developed using a single variable in space and time for independent variables. There are two main categories here: (1) a spatial domain that extends across the eye opening vertically, with or without a moving end, and (2) a shorter domain that is a local model for TF dynamics. We first discuss category (1). First, a number of papers considered a domain that extended across the open eye with fixed ends which typically called post-blink drainage models [91, 92, 93, 94]. These models typically imposed large menisci at the domain ends that drained fluid from the relatively flat interior of the TF, and that drove the dynamics. Notable among these papers is that evaporation was included to find that it could drive thinning to TBU [94], and that a moving end, representing the upstroke of the blink, was treated as a coating flow problem [92]. Both of those aspects would find many applications in later papers. Heat transfer and evaporation were considered in Li and Braun [95], and later treated together with thermal imaging in Dursch et al. [60]. A counterpoint to these papers Berger and Corrsin’s treatment of the Marangoni effect in post blink spreading [14], which provided a basis for using this effect in many later papers.

Using a moving end on a one-dimensional domain to simulate the upper eyelid during a blink has been employed in a number of papers. The first to do this was by Wong et al. [92]; they treated the problem as a coating flow analysis and developed a closed form equation for the thickness of the deposited film as a function of end (lid) speed. Later papers extended this analysis by computing the thickness and other TF variables numerically. These included a combined TF formation and drainage model [96], and then adding surfactant transport on the surface of the TF [97]. Additional papers using the eye opening and interblink together included gravity and reflex tearing [98]; using optimization varying viscosity to simulate and design eyedrops [67]; a detailed look at insoluble surfactant [99]; a (Newtonian) two-layer model of the tear film with insoluble surfactant between them [100]; the effect of (constant) globe curvature on tear film thickness [101]; and a shear-thinning (power law) model for TF thickness on a curved substrate [102]. Some papers also treated the moving end for complete blink cycles. Simple sinusoidal motion was treated in Braun and King-Smith [103], while more realistic motion was treated in Heryudono et al. [104], and insoluble surfactant was added in Zubkov et al. [105]. Heat transfer was later considered with either full [106] or partial [107] blink cycles.

We next discuss category (2) which are local one-dimensional models for TF dynamics. These models are along a short line segment and are designed to be local models of TF flow. The effects of evaporation and the Marangoni effect on TBU have been studied in a number of papers. Evaporation drove all the dynamics in the models in [10, 46, 61], though the treatments were different. In [46], a stationary LL with variable thickness caused increased localized thinning; also included were osmolality transport in the AL and osmosis across the AL/cornea interface. An important result from that work is that diffusion of osmolality out of the region of evaporation prevents osmosis from stopping thinning as it would in spatially uniform models [10, 43]. Simpler evaporation models were used in [10, 61], but they included transport of fluorescein so that they could explain what was seen in TF visualization experiments. Zhong et al. [30] developed a PDE model with one spatial variable that incorporated both mechanisms. Including fluorescein dye transport and fluorescence enabled fitting of models to in vivo fluorescence data within TBU regions to estimate parameters that are not possible to directly measure there at the time of writing [39, 54]. Recently, ODE models with no space dependence have been successfully fit to fluorescence data in small TBU spots and streaks [108]. Those models have been coupled to a neural-network based data extraction system to greatly expand the amount of TBU instances that may be studied [109].

Other effects have been included in local models of TBU [30, 31, 32, 46, 61, 110, 111, 112, 113, 114]. Some models incorporated the effect of soluble mucins in the AL necessitating departure from more standard lubrication models [112, 113]. Localized non-wettability of the ocular surface was modeled in [114]. Other models have provided fundamental analyses on evaporation [115, 116] and instability [117, 118].

LL structure and the model in this paper

The composition and structure of the LL, and their relation to function, are the subject of much ongoing research [77, 119, 120]. The complexity of the composition affects the determination of structure and function, and has led to a variety of perspectives in this area [80, 121]. It is generally thought that the LL has an outermost nonpolar LL with a layer of polar lipids and other surface active components at the LL/AL interface [1, 122]. Measuring all but the most basic properties can be a challenge, however. Microscopy has shown some example forms that the LL may take [23]; the shapes observed there suggest non-Newtonian behavior of the LL [71, 72]. For a number of years, it was very difficult to reproduce significant resistance to evaporation using in vitro meibum or LL films [123, 124], even though resistance to evaporation has been thought to be an important property in vivo [11, 34]. However, recent progress on in vitro films with many components has been made [12, 125]. The barrier function does depend on thickness [34, 126], but clearly there is an effect from internal structure as well [127, 128].

Molecular dynamics simulations, together with Langmuir trough experiments, have also given insight into the structure and function of the LL or model systems for it. Coarse grained models were initially employed [129, 130] and those models suggest that the LL may not be so as well organized as layers of nonpolar lipids atop a polar lipid monolayer. Later, all-atom models have suggested that the specific polar lipids [131, 132] and wax esters [133] may form ordered states to slow evaporation of water from the AL. Molecular dynamics simulations are likely to continue to give important insights into LL structure and function, but they currently can only be solved at small space and time scales. A link to bigger scales has been made via Langmuir trough observations with Brewster angle microscopy, e.g. [131, 132] for equilibrium states. Those insights need to be converted to larger scale dynamic models to address other situations such as TBU. In this paper, we develop a two-layer continuum model that attempts to incorporate some internal structure of the LL that may be expected to produce evaporation resistance [29, 38] that depends on that structure. We use a nematic liquid crystal structure as a model for the LL, which can show birefringence as has been observed for meibum [134].

Adapting two-layer models of tear dynamics [46, 100] to dynamic models of local TBU regions [135, 136] has made use of extensional flow for the LL and shear dominated flow for the AL. These models have a common basis from the approach of Matar et al. [137]. Of particular interest here is including extensional flow of a nematical liquid crystal [138, 139] for the LL with a Newtonian shear-dominated AL layer. That extensional flow modeling used the Ericksen-Leslie equations and multiple scale perturbation methods to reduce the equations to a small PDE system. A detailed derivation and preliminary results for the combined model were developed in Taranchuk and Braun [136]. With this new model, they found dependence of TF dynamics on the director angle θB\theta_B with all other parameters fixed, and we continue this work with a detailed parametric study. We find new ways the effects combine to affect TF dynamics and TBU.

The paper is organized as follows. In § 2 we describe the problem formulation. In § 3 we present our results. These include profiles of the film thicknesses, velocities and pressures that lead to different outcomes depending on the internal orientation of the liquid crystal molecules. Finally, in § 4 we discuss the results and outline our conclusions.

Model

Our two-layer model of the tear film lies on top of the corneal surface, which we take to be flat. It is composed of a Newtonian AL containing various solutes and a liquid crystal LL containing non polar lipids. At the interface of these two fluid layers is a monolayer of surfactant comprising primarily polar lipids. Water in the AL is lost through evaporation, JeJ_e, and supplied from the corneal surface through osmosis, JoJ_o. Figure 1 shows a sketch of the model. In the AL, we solve for the thickness h1h_1 and the osmolarity concentration cc. In the LL we solve for the following quantities: thickness h2h_2, axial velocity u2u_2, angle of the director θ\theta, and surface concentration of surfactant Γ\Gamma.

Schematic of the two-layer tear film model

Figure 1. Schematic of the two-layer tear film model.

We model the AL using the Navier-Stokes equations, and the LL using the Ericksen-Leslie equations for nematic liquid crystals. We use the following scalings, where primes denote dimensional quantities. The coordinates (x′,y′)(x',y') and velocity components (ui′,vi′)(u_i',v_i') correspond to the axial and transverse directions respectively, t′t' is time, and pi′p_i' is pressure. The subscript ii indicates the fluid layer under consideration; i=1i=1 corresponds to the AL, and i=2i=2 corresponds to the LL.

x′=Lx,h1′=ϵLh1,h2′=ϵδLh2,y′=ϵLy,u1′=Uu1,u2′=Uu2,h′=ϵLh,v1′=ϵUv1,v2′=ϵUv2,t′=LUt,p1′=μ1Uϵ2Lp1,p2′=μ2ULp2,Γ′=Γ0Γ,c′=Cc,Jo′=ϵUJo,Je′=ϵρ1UJe.(1)\begin{aligned} x' &= Lx, & h_1' &= \epsilon Lh_1, & h_2' &= \epsilon\delta Lh_2, \\ y' &= \epsilon Ly, & u_1' &= Uu_1, & u_2' &= Uu_2, \\ h' &= \epsilon Lh, & v_1' &= \epsilon Uv_1, & v_2' &= \epsilon Uv_2, \\ t' &= \frac{L}{U}t, & p_1' &= \frac{\mu_1 U}{\epsilon^2 L}p_1, & p_2' &= \frac{\mu_2 U}{L}p_2, \\ \Gamma' &= \Gamma_0\Gamma, & c' &= Cc, \\ J_o' &= \epsilon UJ_o, & J_e' &= \epsilon\rho_1 UJ_e. \tag*{(1)} \end{aligned}

Here the total thickness of the tear film h′=h1′+h2′⟹h=h1+δh2h'=h_1'+h_2' \Longrightarrow h=h_1+\delta h_2. The scalings used in the model, chosen with the dimension of the human eye in mind, are defined in Table 1. The LL is the thinnest layer in the tear film, and has an average thickness of tens to 100 nm during the interblink [13, 140, 141, 142]. The AL is much thicker, averaging about 3 μ\mum [143]. The glycocalyx is less than a micron thick [19, 24]; while it does not appear in the model explicitly, we account for its height by defining TBU as an AL thickness of 0.5 μ\mum or less. We consider a domain of length 2 mm.

ParameterValueUnitsDescriptionReference
L=H1(γ1+γ2)/(μ1E0)1/4L=H_1(\gamma_1+\gamma_2)/(\mu_1E_0)^{1/4}3.1779×10−43.1779\times10^{-4}mLength scale[135]
H1H_13.5×10−63.5\times10^{-6}mAL thickness[143]
H2H_25×10−85\times10^{-8}mLL thickness[140]
V=E0V=E_05.093×10−75.093\times10^{-7}m/sThinning rate[135]
U=V/ϵU=V/\epsilon4.624×10−54.624\times10^{-5}m/sHorizontal velocity[135]
CC300mOsMReference osmolarity[46]
Γ0\Gamma_04×10−74\times10^{-7}mol/m2^2Surfactant concentration[99, 100]

Table 1. Parameters used in dependent variable scalings

The nondimensional quantities that result from scalings are given in Table 2, and the physical parameters are listed in Table 3. We note that while experiments show evidence of liquid crystal structure in the LL of the TF, the parameter values for the LL have not been measured. For liquid crystals, we return to a Newtonian fluid by setting the Leslie viscosities αi=0\alpha_i=0, i=1,⋯ ,6i=1,\cdots,6, i≠4i\ne4, and α4=2μ2\alpha_4=2\mu_2. The dynamic viscosity of the LL, μ2\mu_2, has previously been taken to be 0.1 Pa s [100, 135]. Thus, to match this viscosity in the Newtonian limit, as default values we choose our Leslie viscosities to be three times the properties of 5CB, a liquid crystal that has been relatively well studied near physiological temperatures [144]. We explore a larger range of values in § 3.1.3.

ParameterFormulaValueDescription
ε\varepsilonH1/LH_1/L0.0110Aspect ratio of AL
δ\deltaH2/H1H_2/H_10.0143Lipid to aqueous thickness ratio
γ\gammaε2μ2/μ1\varepsilon^2\mu_2/\mu_10.0091Reduced viscosity ratio
C1C_1ε3(γ1+γ2)/(μ1U)\varepsilon^3(\gamma_1+\gamma_2)/(\mu_1U)1.0001Reduced aqueous capillary number
C2C_2εγ2/(μ2U)\varepsilon\gamma_2/(\mu_2U)43.8371Reduced lipid capillary number
M\mathcal{M}εRT0Γ0/(μ1U)\varepsilon RT_0\Gamma_0/(\mu_1U)187.7687Reduced Marangoni number
E\mathcal{E}E0/(ερ1U)E_0/(\varepsilon\rho_1U)1.0001Evaporation parameter
R0R_0kmH2/(Dk)k_mH_2/(D_k)17.68Evaporative resistance parameter
P\mathcal{P}CPc/(εU)CP_c/(\varepsilon U)0.1355Osmosis parameter
Pe1\mathrm{Pe}_1UL/D1UL/D_19.1842Péclet number for osmolarity diffusion
Pes\mathrm{Pe}_sUL/DsUL/D_s0.4898Péclet number for surfactant diffusion

Table 2. Nondimensional Parameters

ConstantValueUnitsDescriptionReference
ρ1\rho_11000kg/m3\mathrm{kg}/\mathrm{m}^3Density of liquid water[145]
ρ2\rho_2900kg/m3\mathrm{kg}/\mathrm{m}^3Density of lipid[100]
μ1\mu_11.3×10−31.3 \times10^{-3}Pa sDynamic viscosity of water[66, 100]
μ2=α4′/2\mu_2=\alpha'_4/20.0978Pa sDynamic viscosity of lipid[144]
γ1\gamma_10.027N/mSurface tension of water[146]
γ2\gamma_20.018N/mSurface tension of lipid[147]
RR8.3145J/(K mol)Ideal gas constant[145]
D1D_11.6×10−91.6 \times10^{-9}m2/s\mathrm{m}^2/\mathrm{s}Osmolarity diffusion constant[148]
DsD_s3×10−83 \times10^{-8}m2/s\mathrm{m}^2/\mathrm{s}Surfactant diffusion constant[100]
kmk_m0.0182m/sMass transfer coefficient[135]
DkD_k5.136×10−115.136 \times10^{-11}m2/s\mathrm{m}^2/\mathrm{s}Lipid permeability[135]
PcP_c2.3×10−102.3 \times10^{-10}kg/(mOsM m2s)\mathrm{kg}/(\mathrm{mOsM}\,\mathrm{m}^2\mathrm{s})Corneal osmosis coefficient[46]
E0E_05.0928×10−75.0928 \times10^{-7}m/sEvaporative thinning rate[135]
T0T_0308.15KEye surface temperature[46]
α1′\alpha'_1−0.0180-0.0180Pa sLeslie viscosity[144]
α2′\alpha'_2−0.2436-0.2436Pa sLeslie viscosity[144]
α3′\alpha'_3−0.0108-0.0108Pa sLeslie viscosity[144]
α4′\alpha'_40.1956Pa sLeslie viscosity[144]
α5′\alpha'_50.1920Pa sLeslie viscosity[144]
α6′\alpha'_6−0.0624-0.0624Pa sLeslie viscosity[144]
θB\theta_B0−π/20-\pi/2radiansDirector angle[144]

Table 3. Physical Parameters

Governing equations

After nondimensionalizing, we asymptotically expand all of the dependent variables in powers of the small parameter ε\varepsilon. We then collect like powers of ε\varepsilon, and solve to find a closed system of leading order equations. In the process, we find that the director angle is constant at leading order, θ=θB\theta=\theta_B. We also determine the leading order AL pressure p1p_1, and the AL depth averaged axial velocity uˉ1\bar{u}_1. This process is described in more detail in the appendix § 7, and derived thoroughly in Taranchuk and Braun [136]. We find the closed leading order system to be

h10t+(uˉ10h10)x=Jo−Je(2)h_{10t}+(\bar{u}_{10}h_{10})_x=J_o-J_e \tag*{(2)}
h20t+(u20h20)x=0(3)h_{20t} + (u_{20}h_{20})_x = 0 \tag*{(3)}
−A(θB)B(θB)δΥ(u20xh20)x−u20h10=h102(−C1h10xxx−C2Υh0xxx)−C2δΥh20h0xxx+MΓ0x(4)-\frac{A(\theta_B)}{B(\theta_B)}\delta\Upsilon(u_{20x}h_{20})_x-\frac{u_{20}}{h_{10}}=\frac{h_{10}}{2}\left(-C_1h_{10xxx}-C_2\Upsilon h_{0xxx}\right)-C_2\delta\Upsilon h_{20}h_{0xxx}+M\Gamma_{0x} \tag*{(4)}
Γ0t+(Γ0u20)x=1PesΓ0xx(5)\Gamma_{0t}+(\Gamma_0u_{20})_x=\frac{1}{Pe_s}\Gamma_{0xx} \tag*{(5)}
(c0h10)t=1Pel(c0xh10)x−(c0uˉ10h10)x(6)(c_0h_{10})_t=\frac{1}{Pe_l}(c_{0x}h_{10})_x-(c_0\bar{u}_{10}h_{10})_x \tag*{(6)}

where

h0=h10+δh20(7)h_0=h_{10}+\delta h_{20} \tag*{(7)}
Jo=P(c−c0C)(8)J_o=\mathcal{P}\left(c-\frac{c_0}{C}\right) \tag*{(8)}
Je=E1+R(θB)h20(9)J_e=\frac{\mathcal{E}}{1+\mathcal{R}(\theta_B)h_{20}} \tag*{(9)}
R(θB)=R0(0.1+0.9sin⁡θB)(10)\mathcal{R}(\theta_B)=\mathcal{R}_0(0.1+0.9\sin\theta_B) \tag*{(10)}
p10=−ΥC2h0xx−C1h10xx(11)p_{10}=-\Upsilon C_2h_{0xx}-C_1h_{10xx} \tag*{(11)}
uˉ10=−p10xh10212+u202(12)\bar{u}_{10}=-p_{10x}\frac{h_{10}^2}{12}+\frac{u_{20}}{2} \tag*{(12)}

and

A(θB)=2cos⁡(2θB)(α1+α5+α6+4)(α2+α3−α5+α6)+cos⁡(4θB)(α1[α2−α3]+[α2+α3][α5−α6])+α1α2−α1α3−2α1α5−2α1α6−8α1+α2α5+3α2α6+8α2−3α3α5−α3α6−8α3−2α52−4α5α6−16α5−2α62−16α6−32,(13)\begin{aligned} A(\theta_B)={}&2\cos(2\theta_B)(\alpha_1+\alpha_5+\alpha_6+4)(\alpha_2+\alpha_3-\alpha_5+\alpha_6)\\ &+\cos(4\theta_B)(\alpha_1[\alpha_2-\alpha_3]+[\alpha_2+\alpha_3][\alpha_5-\alpha_6])+\alpha_1\alpha_2-\alpha_1\alpha_3-2\alpha_1\alpha_5\\ &-2\alpha_1\alpha_6-8\alpha_1+\alpha_2\alpha_5+3\alpha_2\alpha_6+8\alpha_2-3\alpha_3\alpha_5-\alpha_3\alpha_6-8\alpha_3-2\alpha_5^2\\ &-4\alpha_5\alpha_6-16\alpha_5-2\alpha_6^2-16\alpha_6-32, \tag*{(13)} \end{aligned}
B(θB)=−α1cos⁡(4θB)+α1−2cos⁡(2θB)(α2+α3−α5+α6)−2α2+2α3+2α5+2α6+8.(14)\begin{aligned} B(\theta_B)={}&-\alpha_1\cos(4\theta_B)+\alpha_1-2\cos(2\theta_B)(\alpha_2+\alpha_3-\alpha_5+\alpha_6)\\ &-2\alpha_2+2\alpha_3+2\alpha_5+2\alpha_6+8. \tag*{(14)} \end{aligned}

Equation (2) and (3) represent conservation of mass (of solvent) in the AL and LL, respectively. Equation (4) is the axial force balance in the LL. Equation (5) conserves mass of surfactant, and Equation (6) conserves osmolarity in the AL.

The evaporation function in Equation (9) depends on the evaporative resistance in the LL which is described by Equation (10). We vary the evaporative resistance function R(θB)\mathcal{R}(\theta_B) by changing both R0\mathcal{R}_0 and θB\theta_B. Various director angles are considered throughout the paper, and § 3.1.2 and § 3.2.2 focus on the effect of R0\mathcal{R}_0 for two sets of initial conditions.

Initial and boundary conditions

Evaporation-driven flow

To illustrate flow driven by evaporation, we choose initial conditions in such a way that valley, or thin spot, in the LL drives all the dynamics of the tear film toward TBU. We choose a Gaussian perturbation of the LL

h2(x,0)=1−0.9e−x2/(2σ2),(15)h_2(x,0)=1-0.9e^{-x^2/(2\sigma^2)}, \tag*{(15)}

where 2σ2=1/92\sigma^2=1/9. We take h1=Γ=c=1h_1=\Gamma=c=1 initially. We compute p1(x,0)p_1(x,0) from Equation (21), and solve the discrete version of the axial force balance Equation (22) using the other initial values to obtain u2(x,y,0)u_2(x,y,0). This initial condition corresponds to the Gaussian disturbance described in Stapf et al. [135]; we refer to it as a valley here.

Marangoni effect

To illustrate surfactant gradient driven flow, we choose initial conditions in which all variables are uniform initially (h1=h2=c=1h_1=h_2=c=1), except for a local excess of surfactant in the center of the domain. We compute p1(x,0)=0p_1(x,0)=0 and u2(x,y,0)u_2(x,y,0) as described above. The initial condition for Γ\Gamma is

Γ(x,0)=1+s−s(1+0.5tanh⁡(x−x0xw)+0.5tanh⁡(−x−x0xw)n)(16)\Gamma(x,0)=1+s-s\left(\frac{1+0.5\tanh\left(\frac{x-x_0}{x_w}\right)+0.5\tanh\left(\frac{-x-x_0}{x_w}\right)}{n}\right) \tag*{(16)}

where the height of the region of excess surfactant is s=1.8s=1.8, the inflection point is x0=1x_0=1, and the width of the region is xw=0.36x_w=0.36, and normalization factor n=0.9915n=0.9915.

Combining LL and surfactant perturbations

Having considered perturbations in the LL and surfactant surface concentration individually, we now consider two scenarios which combine perturbations in both. Namely, we consider a bump in h2h_2 combined with a local excess of Γ\Gamma, and a valley in h2h_2 combined with a local excess of Γ\Gamma.

Imaging of the LL has shown instances of regions of thicker lipid [22]. It is reasonable to assume that when the LL is thicker, there is more surfactant. To illustrate this, we consider a thin LL with a central bump and central increased Γ\Gamma in that area. We take h1(x,0)=c(x,0)=1h_1(x,0)=c(x,0)=1, and we calculate p1(x,0)p_1(x,0) and u2(x,0)u_2(x,0). The initial condition for surfactant is Equation (16), and for LL thickness

h2(x,0)=1+s−s(1+0.5tanh⁡(x−x0xw)+0.5tanh⁡(−x−x0xw)n)(17)h_2(x,0)=1+s-s\left(\frac{1+0.5\tanh\left(\frac{x-x_0}{x_w}\right)+0.5\tanh\left(\frac{-x-x_0}{x_w}\right)}{n}\right) \tag*{(17)}

where the height of the bump is s=1s=1, the inflection point is x0=1x_0=1, and the width of the bump is xw=0.36x_w=0.36, and normalization factor n=0.9915n=0.9915. This corresponds dimensionally to a LL with background thickness of 25 nm, and a bump of lipid of height 50 nm in the center of the domain.

We also consider a pathological scenario in which a valley in the LL is combined with a local excess of Γ\Gamma. Keeping all other initial conditions as described above, we modify only Equation (17) by changing s=−0.5s=-0.5, creating a valley in the LL. This corresponds dimensionally to a LL with thickness 50 nm, and a valley of depth 25 nm in the center of the domain.

Boundary conditions

At both boundaries we impose the following no flux boundary conditions,

h1x=h2x=p1x=Γx=cx=0,(18)h_{1x}=h_{2x}=p_{1x}=\Gamma_x=c_x=0, \tag*{(18)}

and specify u2=0u_2=0 for the LL velocity.

Numerical solution

Before solving numerically, we make two change of variables. In order reduce the order of the system, we define p10p_{10} as a dependent variable (Equation (21)) and substitute into Equation (4). We also use the substitution m0=h10c0m_0=h_{10}c_0 in Equation (6), where m0m_0 is the mass. This yields the following closed system which we wish to solve

h10t+(uˉ10h10)x=Jo−Je,(19)h_{10t}+(\bar{u}_{10}h_{10})_x=J_o-J_e, \tag*{(19)}
h20t+(u20h20)x=0,(20)h_{20t}+(u_{20}h_{20})_x=0, \tag*{(20)}
p10=−ΥC2h0xx−C1h10xx,(21)p_{10}=-\Upsilon C_2h_{0xx}-C_1h_{10xx}, \tag*{(21)}
−A(θB)B(θB)δΥ(u20xh20)x−u20h10=12p10x−C2δΥh20h0xxx+MΓ0x,(22)-\frac{A(\theta_B)}{B(\theta_B)}\delta\Upsilon(u_{20x}h_{20})_x-\frac{u_{20}}{h_{10}}=\frac{1}{2}p_{10x}-C_2\delta\Upsilon h_{20}h_{0xxx}+\mathcal{M}\Gamma_{0x}, \tag*{(22)}
Γ0t+(Γ0u20)x=1PesΓ0xx,(23)\Gamma_{0t}+(\Gamma_0u_{20})_x=\frac{1}{\mathrm{Pe}_s}\Gamma_{0xx}, \tag*{(23)}
m0t=1Pe1(m0x−m0h10xh10)x−(m0uˉ10)x,(24)m_{0t}=\frac{1}{\mathrm{Pe}_1}\left(m_{0x}-\frac{m_0h_{10x}}{h_{10}}\right)_x-(m_0\bar{u}_{10})_x, \tag*{(24)}

where

h0=h10+δh20,(25)h_0=h_{10}+\delta h_{20}, \tag*{(25)}
Jo=P(m0h10−1),(26)J_o=\mathcal{P}\left(\frac{m_0}{h_{10}}-1\right), \tag*{(26)}
Je=E1+R(θB)h20,(27)J_e=\frac{\mathcal{E}}{1+\mathcal{R}(\theta_B)h_{20}}, \tag*{(27)}
uˉ10=−p10xh10212+u202.(28)\bar{u}_{10}=-p_{10x}\frac{h_{10}^2}{12}+\frac{u_{20}}{2}. \tag*{(28)}

We then use the method of lines to solve the system numerically subject to the initial and boundary conditions of § 2.2. Spatial derivatives are approximated with second-order finite differences on a uniform grid of 512 points. This results in a system of differential algebraic equations that we solve forward in time in MATLAB (MathWorks, Natick, MA, USA) using ode15s. We found that the solutions were converged using this number of grid points, and that the mass of water and osmolarity were conserved to the tolerances of the computation.

We integrate the system until either of the following happens: (i) a final time for 60 s is reached, or (ii) TBU occurs. We define TBU as the point at which the AL thickness reaches a minimum value of 0.5μm0.5\mu\mathrm{m}, and use event detection to stop the simulation if this happens. When breakup occurs, the TBUT becomes the final time.

Results

Evaporation-driven thinning

As an example of evaporation-driven thinning, we consider the initial conditions described in § 2.2.1 with a director angle of θB=0\theta_B = 0. This choice maximizes the evaporation rate for our chosen evaporation function. Evaporative resistance in the LL is maximized for R(π/2)R(\pi/2); R(0)R(0) reduces that resistance by a factor of 10.

Solutions are shown in Figure 2. The LL changes very little over time, maintaining its original shape with a valley in the center of the domain. The initial velocity in the LL, which ranges between ±50 μm\pm50\ \mu\mathrm{m}, is moderate compared to the dynamics of other scenarios considered later. The thinning in the LL leads to increased evaporation of the AL, which quickly thins, reaching breakup conditions in the center of the domain by 15.9 s. AL pressure begins to drop in the center of the domain; in general, this would help to restore fluid to the area; however TBU occurs because that capillarity-driven contribution is too weak.

Six-panel plot showing solutions for spatially-uniform initial conditions with a Gaussian disturbance in the LL thickness

Figure 2. Solutions for spatially-uniform initial conditions except for a Gaussian disturbance in the LL thickness. The director angle is θB=0\theta_B = 0. This figure illustrates typical dynamics of the dependent variables in space and time for the case of TBU driven by increased evaporation from the valley in the LL.

Osmolarity cc increases over time, peaking in the center of the domain as a result of increased evaporation in that area. By the final time, osmolarity in the center of the domain is 1651 mOsM, more than five times the initial value; and near the ends it is 789 mOsM, over two and a half times the the initial value.

The surface surfactant concentration Γ\Gamma, shown relative to its initial value, builds up slightly in the center of the domain. We explore Marangoni-driven flow in detail in § 3.2; however in this example, the increase in surfactant concentration is not large enough to contribute significantly to the flow.

These solutions are representative of evaporation-driven thinning that leads to breakup before the final time of 60 s. With a larger director angle, evaporation resistance of the LL is increased, and the AL thins more slowly. When θB≲π/17\theta_B \lesssim\pi/17, TBU is not reached within the simulation time of 60 s. Because evaporation is slower, the valley that develops in the AL is less pronounced, and the resulting peak in osmolarity is smaller; see Taranchuk and Braun [136] for more details.

Varying the Marangoni number

We now consider the effect of the Marangoni number on tear film dynamics as the director angle θB\theta_B varies from 0 to π/2\pi/2. Figure 3 shows the extreme values of AL thickness, osmolarity, and LL thickness for three values of the Marangoni number:

Extreme values of AL thickness, osmolarity, and LL thickness for three values of the Marangoni number

Figure 3. Extreme values of AL thickness, osmolarity, and LL thickness at the final time as θB\theta_B is varied from 0 to π/2\pi/2, for three values of the Marangoni number M\mathcal{M}. To the left of the black dots, TBUT occurs before 60 s; to the right, the simulations end at 60 s. The first row corresponds to the initial condition with a valley in the LL described in § 2.2.1. The second row corresponds to the initial condition with a local excess of surfactant described in § 2.2.2.

M\mathcal{M} (the default value of 187.7), M/10\mathcal{M}/10, and 10M10\mathcal{M}. Solutions are also shown for the initial conditions of § 2.2.2, the local excess of surfactant; these are described in further detail below in § 3.2.1.

We see that when M→M/10\mathcal{M} \to\mathcal{M}/10, the Marangoni effect is weakened, and there are no cases of TBU within 60 s. The LL is the most mobile in this case, as healing flow (driven by capillarity) allows the valley in the LL to thicken from its initial minimum of 5 nm, to local minima of 14.7 nm to 33.3 nm, depending on the angle of the director. Osmolarity decreases monotonically from 1323 mOsM to 467 mOsM.

For the default Marangoni number, TBU occurs before 60 s for θB≲π/17\theta_B \lesssim\pi/17. Around this critical director angle the LL has some mobility, with the lipid valley thickening to a maximum of 17.2 nm; however, for θB→π/2\theta_B \to\pi/2, the LL moves very little from its initial profile.

When M→10M\mathcal{M} \to10\mathcal{M}, TBU occurs before 60 s for θB≲π/9\theta_B \lesssim\pi/9. The LL remains tangentially immobile for all angles of the director. Osmolarity decreases over the range of the θB\theta_B, but its decline is slower than for the default Marangoni number.

Varying the evaporative resistance parameter

Figure 4 shows the effect of varying R0\mathcal{R}_0, the evaporative resistance parameter of the LL. The default value R0=17.68\mathcal{R}_0 = 17.68 is based on a nonlinear least squares fit to lipid thickness and permeability measurements performed by Stapf et al. [135]; this is shown with the vertical dashed line. With this level of resistance, only a director angle of θB=0\theta_B = 0 results in a tear breakup time (TBUT) before 60 s. However, if R0R_0 is decreased by a factor of 10, then TBU occurs in under 15 s for all values of the director angle. As the evaporative resistance parameter is increased, the minimum AL thickness tends toward the initial value of 3.5 μm\mu\mathrm{m}.

Tear film breakup time or the minimum thickness of the AL after 60 s as a function of the evaporation resistance parameter

Figure 4. Tear film breakup time or the minimum thickness of the AL after 60 s as a function of the evaporation resistance parameter R0R_0 for the initial condition with a valley in the LL. The black dashed vertical line denotes the default value of R0=17.68R_0 = 17.68.

The effect of liquid crystal viscosities on thinning

We have chosen to set the Leslie viscosities to three times the values of 5CB, a relatively well studied liquid crystal. By doing this, we match the dynamic viscosity of the LL, μ2\mu_2, to previously used value of 0.1 Pa s [100, 135]. However, there is much uncertainty around the values for the Leslie viscosities as they have not been measured for the LL, and using a scale factor of 3 is only one possible choice. Thus, in Figure 5, we consider the effect of a wide range of scalings on the minimum values for both the aqueous and LL. Note that because we are scaling the dimensional quantities, in the nondimensional form we are effectively scaling μ2\mu_2. Here we reduce the Marangoni number by a factor of 10 (M→M/10\mathcal{M} \to\mathcal{M}/10) in order to isolate the effect of the liquid crystal viscosities on tear film thinning and breakup.

Plots of minimum thickness and TBUT versus viscosity scale factor for different values of $\theta_B$

Figure 5. Minimum thickness of the AL after 60 s or TBUT as a function of the scale factor applied to the 5CB liquid crystal parameters. To the left of the black dots, TBUT occurs before 60 s; to the right, the simulations end at 60 s.

For values of μ2\mu_2 that are increased by less than a factor of 1000, there is little change in either the minimum AL thickness or LL thickness at the final time. For values of μ2\mu_2 more than a factor of 1000 larger than default, we begin to see breakup within 60 s for small values of θB\theta_B. However, for larger director angles, the AL is fairly unaffected. On the other hand, for small values of μ2\mu_2 we observe a much more mobile LL for all orientations of the director. As μ2\mu_2 increases, the LL becomes immobile, and the minimum thickness remains close to the initial value of 5 nm.

Figure 6 shows solutions at the final time when μ2\mu_2 is increased by a factor of 2000. At this value, we see the most variation in LL thickness with respect to the director angle. For clarity, we omit θB=π/18\theta_B = \pi/18 and π/4\pi/4. When θB=π/2\theta_B = \pi/2, the minimum LL thickness is just over 12 nm, while when θB=0\theta_B = 0, the effects of capillarity result in healing flow, which increases the minimum thickness to 31.7 nm. For the intermediate value of θB=π/12\theta_B = \pi/12, the LL experiences some healing flow with a minimum of 24.1 nm, but mobility is reduced.

Solutions for the LL at 60 s

Figure 6. Solutions for the LL at 60 s when the initial condition has a valley in the LL and μ2→2000μ2\mu_2 \to2000\mu_2, and M→M/10\mathcal{M} \to\mathcal{M}/10.

It is natural to explore an alternative scenario: keeping a scaling of three on α4′\alpha'_4, which maintains μ2=0.1\mu_2 = 0.1 Pa s, while scaling the other Leslie viscosities, αi′\alpha'_i, i=1,…,6i = 1,\ldots,6, i≠4i \ne4. By doing this, we are changing the coefficient A/BA/B on the extensional term in Equation (4). Note that as αi′→0\alpha'_i \to0, i=1,…,6i = 1,\ldots,6, i≠4i \ne4, we return to the Newtonian limit, where A/B=4A/B = 4. However, we find that if we increase the scaling too much (more than a factor of 16), the coefficient A/BA/B becomes positive for some values of θB\theta_B, making the problem ill-posed. In addition, the effect of the scalings in this feasible range is negligible for all variables. If we look at the coefficient A/BA/B for scalings, shown in Figure 7, we see that the change on the coefficient is very little, even with the maximum scale factor of 16.

The coefficient $A/B$ as $\theta_B$ varies

Figure 7. The coefficient A/BA/B as θB\theta_B varies for scalings of 3, 10, and 16 on the Leslie viscosities αi\alpha_i, i=1,⋯ ,6i=1,\cdots,6, i≠4i\ne4 for the liquid crystal 5CB.

Marangoni-driven thinning

We now turn to the initial conditions of § 2.2.2, in which concentration gradients drive the tear film dynamics. The Marangoni effect results in rapid tangential flow; thus we consider solutions on a short timescale, shown in Figure 8, and a longer timescale, shown in Figure 9. Figure 8 shows the evolution of the variables over the first 0.06 s, when θB=π/2\theta_B = \pi/2. These solutions are representative of the dynamics of tear film thinning due to surfactant concentration gradients regardless of the orientation of the liquid crystal molecules; the director angle θB\theta_B has little effect on the short timescale. The large surfactant gradients immediately cause rapid thinning in both the lipid and aqueous layers. As the local excess of surfactant flattens, the Marangoni effect concludes. The LL thins dramatically, with the valley thickness roughly half of the initial value of 50 nm. Fluid is pulled away from the excess surfactant, and builds up on the edges of the domain. The pattern of thinning is very similar in the AL, although to a lesser degree. During this initial redistribution of fluid, the LL fluid velocity is very large in magnitude (around 3×105 μm/min3 \times10^{5}\,\mu\mathrm{m}/\mathrm{min}), but quickly slows to the range of ±5 μm/min\pm5\,\mu\mathrm{m}/\mathrm{min} throughout the domain (not shown). Similarly, AL pressure (not shown) drops in magnitude from a range of ±3\pm3 Pa during the first 1 s, to ±1\pm1 Pa, and thinning continues over a longer timescale.

Solutions on a short time scale for spatially-uniform initial conditions except for a local excess of surfactant

Figure 8. Solutions on a short time scale for spatially-uniform initial conditions except for a local excess of surfactant. The director angle is θB=π/2\theta_B = \pi/2. This figure illustrates typical dynamics of the dependent variables in space and time. The surface gradients of surfactant drive all the dynamics.

Solutions at the final time for spatially-uniform initial conditions except for a local excess of surfactant

Figure 9. Solutions at the final time for spatially-uniform initial conditions except for a local excess of surfactant. When θB=0\theta_B = 0, breakup occurs at 16.6 s; for all other values of θB\theta_B the final time is 60 s. The LL profiles (upper right) are nearly identical for all θB\theta_B.

Figure 9 shows solutions at the final time for five values of θB\theta_B ranging from 0 to π/2\pi/2. Breakup conditions are reached when θB=0\theta_B = 0 by 16.6 s; for all other values of θB\theta_B, the final time of 60 s is reached. The surfactant concentration remains effectively uniform around 1.58 times the original background concentration, and over the longer timescale it is the evaporative resistance in the LL that influence thinning dynamics. When θB=π/2\theta_B = \pi/2 or π/4\pi/4 thinning becomes more localized in the center of the domain, while smaller values of θB\theta_B maintain a wider valley of thinning through time. For larger values of θB\theta_B, we observe that negative pressure in the center of the AL domain pulls fluid in, and helps to reduce thinning. This results in an overall thicker AL at the final time. Arrows in the velocity and pressure plots of Figure 9 point to the difference in behavior when θB=π/4\theta_B = \pi/4 or π/2\pi/2 as compared to smaller director angles.

On the other hand, the LL profiles are indistinguishable from each other at the final time and look very similar to the solution at 0.06 s shown in Figure 8. Virtually all of the movement in the LL happens within the first second; this immobilization of the fluid surface due to the Marangoni effect has been studied in other contexts such as vertical drainage in Braun et al. [149] and remobilization of droplet surfaces in Stebe and Maldarelli [150].

Osmolarity in the AL increases for all values of θB\theta_B over the original value of 300 mOsM; however, it is most pronounced when θB=0\theta_B = 0. While the solutions in the AL are not dissimilar for θB=0\theta_B = 0 and θB=π/18\theta_B = \pi/18, the osmolarity is very different. When θB=0\theta_B = 0, peak osmolarity reaches nearly 1400 mOsM by 16.6 s, more than one and a half times the peak osmolarity at 60 s when θB=π/18\theta_B = \pi/18. When θB=0\theta_B = 0, thinning and TBU happen so rapidly that the solutes do not have enough time to diffuse through the AL.

Varying the Marangoni number

The second row of Figure 3 shows extreme values for AL thickness, osmolarity, and LL thickness as the director angle varies from 0 to π/2\pi/2 for three values of the Marangoni number: the default M\mathcal{M} (187.7), M/10\mathcal{M}/10, and 10M10\mathcal{M}. In general, compared to the initial condition with a valley in the LL which are discussed in § 3.1.1, varying the Marangoni number has a much smaller effect on the variables. This is because with localized excess surfactant in the initial condition, the Marangoni effect is already driving dynamics, and our default Marangoni number is strong enough that increasing or decreasing by a factor of 10 does not have a dramatic effect.

For all three values of the Marangoni number, TBU occurs for θB<π/50\theta_B < \pi/50 when the initial condition has a local excess of surfactant. For larger director angles, there is no breakup within 60 s, but the AL thins much more than in the case with the lipid valley, with minimum AL thickness ranging from 1.62 to 1.84 μ\mum when θB=π/2\theta_B = \pi/2.

After the initial rapid redistribution of fluid due to the Marangoni effect, the valley formed in the LL has a minimum value of 28 nm when θB=0\theta_B = 0 for all three cases. For M\mathcal{M} and 10M10\mathcal{M} the minimum remains the same for all values of the director. For M/10\mathcal{M}/10, the minimum increases slightly as the director angle increases, with a minimum value of 31.6 nm when θB=π/2\theta_B = \pi/2. Osmolarity between the three cases is very similar, with a maximum difference of 28 mOsM; in fact, the osmolarity curve for 10M10\mathcal{M} is indistinguishable from that of M\mathcal{M} in Figure 3.

Varying the evaporative resistance parameter

We now explore the effect of the evaporative resistance parameter R0R_0. Figure 10 shows the TBU conditions as R0R_0 varies for various orientations of the director. A small evaporative resistance parameter results in rapid thinning of the AL with breakup just under 5 s for all values of θB\theta_B. As R0R_0 increases, breakup time increases, until eventually the final time of 60 s is reached before breakup occurs. The value of R0R_0 at which this occurs varies from 2.5 for θB=π/2\theta_B = \pi/2 to 25.1 for θB=0\theta_B = 0. The default value, denoted by the dashed vertical line, falls into this range where thinning may result in breakup before 60 s, depending on the director angle. See Figure 9 for solutions at the final time. As R0R_0 becomes very large, evaporation is diminished, and the minimum value tends toward 3.1 μ\mum for all values of θB\theta_B.

Tear film breakup time or the minimum thickness of the AL after 60 s as a function of the evaporation resistance parameter

Figure 10. Tear film breakup time or the minimum thickness of the AL after 60 s as a function of the evaporation resistance parameter R0R_0 for the initial condition that has a local excess of surfactant. The black dashed vertical line denotes the default value of R0=17.68R_0 = 17.68.

Figure 11 shows the relationship between the evaporative resistance parameter and the angle of the director for the two sets of initial conditions. The value of R0R_0 which results in breakup conditions at 60 s is shown in red for θB=0\theta_B = 0, π/18\pi/18, π/12\pi/12, π/4\pi/4, and π/2\pi/2. The circles represent the solution from the evaporation-driven flow caused by a valley in the LL; the crosses represents solutions from flow driven by the Marangoni effect caused by localized excess surfactant. The black lines are rational functions fit to the respective data delineating whether or not breakup conditions are reached by the final time of 60 s. If the evaporative resistance parameter is decreased by a factor of 10 from the default value of 17.68, then TBU will occur for all angles of the director for both sets of initial conditions, although TBUT will vary.

Comparison of the $R_0$ value resulting in TBU at 60 s for two sets of initial conditions

Figure 11. A comparison of the R0R_0 value which results in TBU at 60 s for two sets of initial conditions: one describing evaporation-driven flow (red circles), and one describing flow driven by the Marangoni effect (red crosses), as a function of the director angle of the liquid crystal molecules in the LL. The black lines are a fit to the respective data. To the left of each black line, TBU occurs before 60 s; to the right of each black line, TBU may occur after 60 s.

If we compare the two curves, we see that the R0R_0 which results in TBU at 60 s for the case of evaporation-driven thinning is roughly double that of the case of Marangoni-driven flow. In the evaporative case, when θB=0\theta_B = 0, the evaporative resistance parameter R0=50R_0 = 50, whereas for the Marangoni-driven thinning, R0=25R_0 = 25. When θB=π/2\theta_B = \pi/2, R0R_0 is 5 and 2.5, respectively. This trend roughly holds for all values of the director angle. In general, this is expected behavior; the LL acts as a barrier to evaporation in part due to its thickness. In the first set of initial conditions, a deficiency in the LL leads to increased evaporation. Thus, it cannot tolerate an evaporation resistance parameter value of any less than 50 without TBU occurring before 60 s when θB=0\theta_B = 0. On the other hand, in the Marangoni-driven flow, the LL begins with a uniform thickness of 50 nm which helps to prevent TBU even though other aspects of the barrier may be compromised.

Reducing initial lipid layer thickness

The thickness of the LL ranges between 20 - 120 nm, with more recent measurements favoring a thinner LL [140, 141, 142]. We halve the initial LL thickness from 50 nm (a “normal” tear film thickness), to 25 nm (a “thin” thickness). Figure 12 shows solutions for the AL and osmolarity at the final time for a few values of θB\theta_B. The solutions can be compared with those of Figure 9 which show solutions for a 50 nm thick initial LL. Qualitatively, solutions for large values of θB\theta_B are similar, although the AL is thinner, especially at its minimum. However, for θB=π/12\theta_B = \pi/12 or π/18\pi/18, there are two minima, located symmetrically about the center. For θB=0\theta_B = 0, the region of thinning is steeper and more narrow than when the LL is thicker. In all cases, the solutions for osmolarity show higher concentrations with a thinner LL.

Solutions at the final time for spatially-uniform initial conditions

Figure 12. Solutions at the final time for spatially-uniform initial conditions except for a local excess of surfactant. The initial thickness of the LL is 25 nm (half the default value). When θB=0\theta_B = 0, breakup occurs at 8.2 s; when θB=π/18\theta_B = \pi/18, breakup occurs at 54.8 s. For all other values of θB\theta_B, the final time is 60 s.

Table 4 shows that a 50% decrease in LL thickness results in a 40.1% decrease in the minimum AL thickness when θB=π/4\theta_B = \pi/4; for other values of the director, the percent decrease is even less. The factor increase in peak osmolarity over the ideal osmolarity is also shown. These numbers are highest when TBU occurs; however, with a thinner LL, peak osmolarity is well over twice the ideal value for all angles of the director.

θB\theta_B\multicolumn{3}{c}{h1min⁡h_{1\min}}\multicolumn{4}{c}{cmax⁡c_{\max}}
50 nm (TBUT)25 nm (TBUT)% decrease50 nmfactor ↑\uparrow25 nmfactor ↑\uparrow
00.5 (16.6 s)0.5 (8.2 s)013984.6614574.86
π/18\pi/180.66260.5 (54.8 s)24.59223.0712734.24
π/12\pi/120.75380.552826.78092.7011343.78
π/4\pi/41.33610.801040.15461.827682.56
π/2\pi/21.64720.992639.74781.596592.20

Table 4. Comparison of minimum AL thickness and maximum osmolarity when the initial LL thickness is reduced from 50 nm to 25 nm. The factor increase in peak osmolarity is compared to the ideal osmolarity of 300 mOsM.

Figure 13 shows the effect of varying the director angle by plotting parametric changes to the minimum value of the AL thickness and the maximum value of osmolarity at the final time as θB\theta_B varies from 0 to π/2\pi/2 for both a normal and thin LL. When the LL thickness is 50 nm, TBU occurs before 60 s only for values of θB<π/50\theta_B < \pi/50; when the LL thickness is 25 nm, TBU occurs for θB<π/14\theta_B < \pi/14. The black dots on the plot mark this change. Minimum LL thickness as θB\theta_B varies is not shown, as changes are minimal. When the initial LL thickness is 50 nm, minimum values hovers near 28 nm; when the initial thickness is 25 nm, the minimum values remain around 14 nm for all values of the director angle.

Extreme values of AL thickness, osmolarity, and LL thickness at the end of each simulation

Figure 13. Extreme values of AL thickness, osmolarity, and LL thickness at the end of each simulation as θB\theta_B is varied from 0 to π/2\pi/2. To the left of the black dot, TBUT occurs before 60 s; to the right of this dot, the simulations end at 60 s.

Combining lipid layer and surfactant perturbations

We now compare the two scenarios which combine perturbations in both the LL thickness and the surfactant concentration. The first combination is a bump in h2h_2 with a local excess of Γ\Gamma, and the second is a valley in h2h_2 with a local excess of Γ\Gamma. Figure 14 compares solutions for AL and LL thickness, as well as osmolarity for these two cases with θB=π/2\theta_B = \pi/2. While neither case reaches breakup for this angle of the director, thinning in the AL is more pronounced when there is a valley in the LL, with a minimum value of 0.94 μm0.94\,\mu\mathrm{m}, as compared to 1.66 μm1.66\,\mu\mathrm{m} when the LL contains a bump. The LL behavior is similar in both cases; it undergoes rapid redistribution of fluid initially, and then remains tangentially immobile for the remainder of the time. When the LL contains a bump, this extra fluid is quickly redistributed throughout the domain; an area of thinning results, with the minimum thickness of 28.46 nm28.46\,\mathrm{nm}. When the LL contains a valley, the Marangoni effect magnifies this profile resulting in a solution similar to those of Figure 9, but with a deeper valley: the depth is now 14.15 nm14.15\,\mathrm{nm} versus 28.1 nm28.1\,\mathrm{nm} (when θB=π/2\theta_B = \pi/2).

Six-panel plot showing solutions for a bump and a valley in $h_2$ combined with a local excess of $\Gamma$

Figure 14. Left column: Solutions when θB=π/2\theta_B = \pi/2 for a bump h2h_2 combined with a local excess of Γ\Gamma. Right column: Solutions when θB=π/2\theta_B = \pi/2 for a valley in h2h_2 combined with a local excess of Γ\Gamma. In both cases Γ(x,0)\Gamma(x, 0) is given by Equation (16). The different h2(x,0)h_2(x, 0) appears in the middle row.

Osmolarity in both cases reflects the respective changes in the AL. When the LL contains a valley, the AL thinning is more pronounced, and thus osmolarity reaches a peak of 650 mOsM, with little increase in concentration at the edges of the domain. When the LL contains a bump, the AL thins to a lesser degree, and the osmolarity steadily rises throughout the domain, with a smaller peak in the center. Peak osmolarity at the final time reaches 485 mOsM. In all scenarios that contain a local excess of surfactant initially, the Marangoni effect results in a rapid flattening of the surfactant concentration to a more or less uniform value of 1.58.

We can consider the influence of the director angle, by observing the final time solutions for various angles as shown in Figure 15. When there is a bump in the LL, we see more variation resulting from changes in the director angle. When θB=0\theta_B = 0, TBU occurs at 46.8 s, but no other angles of the director result in breakup. When there is a valley in the LL, the director angle has less of an impact because rapid thinning in the AL leads to TBU for θB=0\theta_B = 0, π/18\pi/18, and π/12\pi/12. Breakup times are detailed further in Table 5 which is discussed more below. Osmolarity at the final time is strongly influenced by the director angle. When there is a bump in the LL, peak osmolarity is almost three times higher for θB=0\theta_B = 0 (1409 mOsM) than for θB=π/2\theta_B = \pi/2 (485 mOsM). The difference is just over double when there is a valley in the LL (1442 versus 650 mOsM).

Solutions at the final time for a bump in $h_2$ combined with a local excess of $\Gamma$, and for a valley in $h_2$ combined with a local excess of $\Gamma$

Figure 15. Left column: Solutions at the final time for a bump in h2h_2 combined with a local excess of Γ\Gamma. Right column: Solutions at the final time for a valley in h2h_2 combined with a local excess of Γ\Gamma.

LL max/min (nm)00π/18\pi/18π/12\pi/12π/4\pi/4π/2\pi/2
50/258.229.648.4--
40/207.415.032.2--
30/156.710.714.3--

Table 5. TBU time comparison in seconds for various initial LL thicknesses when a valley in the LL is combined with a local excess of surfactant.

Thus far we have considered a valley in the LL with depth 25 nm in a 50 nm LL. We now vary this profile keeping a valley of half thickness. Table 5 compares the TBUT for three different LL background thicknesses: 30, 40, and 50 nm. We see that the dynamics are not sensitive to LL thickness when θB=0\theta_B = 0. This is because evaporative resistance is at its lowest for this angle of the director; for all of the scenarios we have considered, θB=0\theta_B = 0 results in TBU before 60 s. LL thickness has a stronger effect for intermediate angles of the director. For example, when θB=π/18\theta_B = \pi/18, decreasing the LL thickness by 20%, from 50 nm to 40 nm, results in TBU in half the time. When θB\theta_B is large enough, TBU does not occur; for all three LL thicknesses TBU does not occur for θB=π/4\theta_B = \pi/4 or π/2\pi/2.

Discussion and conclusion

In this paper we explore a new model for the tear film of the eye that was derived in Taranchuk and Braun [136]. This two-layer model combines a Newtonian AL with a nematic liquid crystal LL, and links molecular orientation of the LL with evaporative resistance that depends on the director angle. The LL acts as a barrier to evaporation of the AL, and thickness alone does not explain its role in preventing evaporation [127, 128]. By representing the LL as nematic liquid crystal, we are able to consider possible connections between the structure and composition of the LL and its affect on tear film thinning and breakup.

We examined two key drivers of TBU: evaporation and the Marangoni effect. We consider several scenarios which may lead to TBU. The first is a narrow valley in the LL, which leads to thinning of the AL due to increased evaporation. The second is a local excess of surfactant on the AL/LL interface, for which we see thinning on two timescales. Initially, surfactant gradients lead to rapid outward flow in both the AL and LL. The surfactant concentration levels out within the first second, and then evaporation takes over, continuing tear film thinning over a longer timescale of 60 s. Finally, we considered combinations of perturbations in both the LL and surface surfactant; one in which there is a bump in the LL combined with a local excess of surfactant, and one in which there is a valley in the LL combined with a local excess of surfactant. This final case depicts a worst case scenario, where both excess surfactant and a deficiency in the LL drive thinning, and breakup conditions are reached for θB≲π/12\theta_B \lesssim\pi/12.

The evaporative resistance parameter R0R_0 directly influences how quickly the AL thins. However, as shown in Figure 11, the minimum value of R0R_0 that the system can tolerate while withstanding breakup also depends on the angle of the director, θB\theta_B, and the initial condition for the LL. The evaporation function, given in Equations (9) and (10) and derived in A.3, shows one way these quantities may be related. While we have chosen to introduce director angle dependence in the evaporation function [29, 38, 151], we also see that the director angle has an effect on flow dynamics. Small angles of θB\theta_B cause the LL to become more mobile, which results from liquid crystal properties in the LL.

In addition, we find that the default Marangoni number has a strong effect on dynamics, regardless of the initial conditions or other parameter values used. Thus, in order to observe the effect of the liquid crystal parameters, especially the Leslie viscosities, we reduce the Marangoni number by a factor of ten. In the absence of any measurements in the eye, we use the properties of 5CB as a starting point for the Leslie viscosities of the liquid crystal lipid layer. One unexpected finding is that varying the Leslie viscosities from the 5CB values has little effect on the dynamics. Not only that, but we are constrained in how much we can vary them, at least when using constant scalings of 5CB. On the other hand, the other main liquid crystal parameter θB\theta_B, which describes the orientation of the molecules, affects both evaporation (by design), and flow. This is especially true for small angles.

Rather than varying all of the Leslie viscosities, we focused on varying α4′\alpha'_4, which controls the dynamic viscosity of the LL (because μ2=α4′/2\mu_2 = \alpha'_4/2). These results are very similar to the Newtonian case, explored in Stapf et al. [135], in that it takes a very large increase in μ2\mu_2 to effect the dynamics of the AL. The director angle does have an effect on the mobility of the LL, until μ2\mu_2 become very large (on the order of 10510^5). After this point, the LL becomes virtually tangentially immobile for all angles of the director.

Finally, we consider two initial conditions which combine perturbations in both the LL and the surface surfactant. The first is an undesirable case where a valley of lipid is combined with a local excess of surfactant. As expected, we see more breakup cases; for all angles of θB≲10\theta_B \lesssim10, TBU occurs within 60 s as a result of both the Marangoni effect and increased evaporation due to the deficiency in the LL.

Alternatively, we consider an initial condition which combines a bump in the LL with a local excess of surfactant. This scenario corresponds to Section 5.6 in Bruna and Breward [100], where in a limit h2h_2 and Γ\Gamma can both be governed by the same equation. While we do not take that limit here, we model an initial condition in which the LL follows the surfactant. This area of increased thickness in the LL seem to be protective by creating a buffer against the Marangoni effect, and thus helping to prevent excessive thinning in the AL. Imaging of the LL has shown areas of varying thickness on occasion [22]. This type of initial condition could resemble this scenario, but more work is needed to link them.

Choudhury et al. [114] modeled TBU driven by dewetting due loss of mucins in the glycocalyx using a single layer model. They consider 119 s to be a normal time to breakup in a healthy eye, and their simulation resulted in TBU as short as 3.3 s when there is mucin loss over a third of the corneal surface. Dey et al. [112, 113] studied the same mechanism for TBU, but used a continuous-viscosity model which combined the mucus and aqueous layers. Their model predicted breakup in 2.4 to 135 s for scenarios representing pathological to healthy conditions, respectively. in vivo measurements for subjects with dry eye or other eye conditions have shown TBU times well under 10 s. In healthy subjects, TBU times may be longer, but would generally occur in less than two minutes [33, 37, 126].

While our model focuses on different mechanisms for thinning and breakup, TBU times from our simulations fall within a similar range as previous models and experiments. We consider that a healthy eye has high evaporative resistance, corresponding to a large director angle. For the scenarios we considered, a director angle of θB=π/4\theta_B = \pi/4, or π/2\pi/2 did not result in breakup within 60 s. We did not compute time to TBU; however, at the end of 60 s, minimum AL thicknesses were well above the threshold value of 0.5 μ\mum for these larger director angles. For small angles of the director in which evaporative resistance was low and there was a deficiency in the LL or an excess of surfactant, our model predicted breakup in as little as 6 to 10 s. These predicted breakup times are not as rapid as the shortest TBU times observed in eyes [33, 126].

Thinning and breakup due to a glob surfactant was studied in Zhong et al. [30, 111]. The glob consisted of an interval of fixed concentration as well as a part that was transported outside this interval. This area of fixed concentration, which acted as a source, was varied in size, and found to cause TBU either under or at the edge of the glob depending on size. TBU occurred in < 1 s to 15 s, and it was noted that evaporation aided TBU. Our results have central breakup and evaporation aligns with the Marangoni effect to promote TBU. Although our TBU time is on the longer end of the range they predicted, our treatment of surfactant is simpler. We conserve surfactant, and the local excess of surfactant does not act as a source; nor did we experiment with larger or smaller concentrations.

Thinning due to evaporation has been modeled previously both with a fixed evaporation distribution [61], and with a static LL [46]. Alternatively, Luke et al. [54] developed an evaporation-driven model that was fit to experimental data. We have taken a mechanistic approach, and incorporated LL thickness, permeability, and structure in our evaporation function.

We have studied a variety of initial conditions and explored the effects of several key parameters on tear film dynamics when modeling the LL as nematic liquid crystal. Now that we have a sense of the key features and limitations of this model, future work could include a moving end to imitate the eyelid during a blink. Tear breakup is often related to the LL pattern [10, 126]; more fidelity would require a moving domain [97]. In addition, it would be valuable to determine more rheological properties for this kind of model [70, 71, 72].

Acknowledgements

This material is based upon work supported by the University of Delaware Graduate College through the Doctoral Fellowship for Excellence (MJT). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors. RJB is grateful for SHD’s mentoring and many examples of the right way to do things. Remembering how he got his unofficial middle name, we have joined words wherever a possibility of using a hyphen has occurred in this paper.

Declarations

Competing interests

The authors have no competing interests to declare that are relevant to the content of this article.

Dimensional equations

We give the dimensional equations (after [136]) and a very brief outline of the derivation.

Governing equations

Inside the AL, 0<y′<h1′0 < y^{\prime} < h^{\prime}_{1}, we have conservation of mass and momentum using the Navier-Stokes equations. The advection-diffusion equation governs osmolarity transport within the AL. The equations in dimensional form with a prime denoting a dimensional quantity, respectively, are:

∇′⋅u1′=0,(29)\nabla^{\prime} \cdot\mathbf{u}^{\prime}_{1} = 0, \tag*{(29)}
ρ1(u1t′+u1′⋅∇′u1′)=−∇′p1′+μ1Δ′u1′,(30)\rho_{1}\left(\mathbf{u}^{\prime}_{1t}+\mathbf{u}^{\prime}_{1}\cdot\nabla^{\prime}\mathbf{u}^{\prime}_{1}\right)=-\nabla^{\prime}p^{\prime}_{1}+\mu_{1}\Delta^{\prime}\mathbf{u}^{\prime}_{1}, \tag*{(30)}
ct′+u1′⋅∇′c′−D1Δ′c′=0(31)c'_{t}+ \mathbf{u}'_{1}\cdot\nabla'c' -D_{1}\Delta'c'=0 \tag*{(31)}

where the divergence of u1′\mathbf{u}'_{1} is ∇′⋅u1′=u1x′′+v1y′′\nabla'\cdot\mathbf{u}'_{1}=u'_{1x'}+v'_{1y'}, the gradient is

∇′u1′=[u1x′′v1x′′u1y′′v1y′′],\nabla'\mathbf{u}'_{1}= \begin{bmatrix} u'_{1x'} & v'_{1x'}\\ u'_{1y'} & v'_{1y'} \end{bmatrix},

and the Laplacian of the first component of u1′\mathbf{u}'_{1} is Δ′u1′=u1x′x′′+u1y′y′′\Delta'u'_{1}=u'_{1x'x'}+u'_{1y'y'}.

Inside the LL, h1′<y′<h′h'_{1}<y'<h', we have conservation of energy, mass, and momentum using the Ericksen-Leslie equations for liquid crystals.

∂∂x′(∂W′∂θx′)+∂∂y′(∂W′∂θy′)−∂W′∂θ+g~x′∂nx∂θ+g~y′∂ny∂θ=0(32)\frac{\partial}{\partial x'}\left(\frac{\partial W'}{\partial\theta_{x'}}\right)+\frac{\partial}{\partial y'}\left(\frac{\partial W'}{\partial\theta_{y'}}\right)-\frac{\partial W'}{\partial\theta}+\tilde{g}'_{x}\frac{\partial n_{x}}{\partial\theta}+\tilde{g}'_{y}\frac{\partial n_{y}}{\partial\theta}=0 \tag*{(32)}
−∂π′∂x′+g~x′∂nx∂x′+g~y′∂ny∂x′+∂τ~xy′∂y′+∂τ~xx′∂x′=0(33)-\frac{\partial\pi'}{\partial x'}+\tilde{g}'_{x}\frac{\partial n_{x}}{\partial x'}+\tilde{g}'_{y}\frac{\partial n_{y}}{\partial x'}+\frac{\partial\tilde{\tau}'_{xy}}{\partial y'}+\frac{\partial\tilde{\tau}'_{xx}}{\partial x'}=0 \tag*{(33)}
−∂π′∂y′+g~x′∂nx∂y′+g~y′∂ny∂y′+∂τ~yx′∂x′+∂τ~yy′∂y′=0(34)-\frac{\partial\pi'}{\partial y'}+\tilde{g}'_{x}\frac{\partial n_{x}}{\partial y'}+\tilde{g}'_{y}\frac{\partial n_{y}}{\partial y'}+\frac{\partial\tilde{\tau}'_{yx}}{\partial x'}+\frac{\partial\tilde{\tau}'_{yy}}{\partial y'}=0 \tag*{(34)}
∂u2′∂x′+∂v2′∂y′=0(35)\frac{\partial u'_{2}}{\partial x'}+\frac{\partial v'_{2}}{\partial y'}=0 \tag*{(35)}

where, working in two dimensions,

g~i′=−γrNi′−γteik′nk,eij′=12(∂ui′∂xj′+∂uj′∂xi′)(36)\tilde{g}'_{i}=-\gamma_{r}N'_{i}-\gamma_{t}e'_{ik}n_{k},\qquad e'_{ij}=\frac{1}{2}\left(\frac{\partial u'_{i}}{\partial x'_{j}}+\frac{\partial u'_{j}}{\partial x'_{i}}\right) \tag*{(36)}
Ni′=n˙i′−ωik′nk,ωij′=12(∂ui′∂xj′−∂uj′∂xi′),π′=p2′+W′(37)N'_{i}=\dot{n}'_{i}-\omega'_{ik}n_{k},\qquad \omega'_{ij}=\frac{1}{2}\left(\frac{\partial u'_{i}}{\partial x'_{j}}-\frac{\partial u'_{j}}{\partial x'_{i}}\right),\qquad \pi'=p'_{2}+W' \tag*{(37)}
W′=12[K1(∇′⋅n)2+K3((n⋅∇′)n)⋅((n⋅∇′)n)](38)W'=\frac{1}{2}\left[K_{1}(\nabla'\cdot\mathbf{n})^{2}+K_{3}((\mathbf{n}\cdot\nabla')\mathbf{n})\cdot((\mathbf{n}\cdot\nabla')\mathbf{n})\right] \tag*{(38)}
τ~ij′=α1′nknpekp′ninj+α2′Ni′nj+α3′Nj′ni+α4′eij′+α5′eik′nknj+α6′ejk′nkni.(39)\tilde{\tau}'_{ij}=\alpha'_{1}n_{k}n_{p}e'_{kp}n_{i}n_{j}+\alpha'_{2}N'_{i}n_{j}+\alpha'_{3}N'_{j}n_{i}+\alpha'_{4}e'_{ij}+\alpha'_{5}e'_{ik}n_{k}n_{j}+\alpha'_{6}e'_{jk}n_{k}n_{i}. \tag*{(39)}

We use summation notation here as it is standard when working with the Ericksen-Leslie equations, with i,j,k=1,2i,j,k=1,2, and n˙i\dot{n}_{i} denotes the convective derivative of the iith component of n\mathbf{n}. We further assume that the elastic constants are equal; that is K=K1=K3K=K_{1}=K_{3}. These liquid crystal quantities are defined further in Table A1.

QuantityDescription
n=(sin⁡θ,cos⁡θ)\mathbf{n}=(\sin\theta,\cos\theta)director field
θ(x′,y′,t′)\theta(x',y',t')angle the director angle makes with the yy-axis
p2′p'_2pressure in LL
W′W'bulk energy density
π′=p2′+W′\pi'=p'_2+W'modified pressure
gij′g'_{ij}viscous dissipation
eij′e'_{ij}rate of strain tensor
Nij′N'_{ij}co-rotational time flux of the director n\mathbf{n}
ωij′\omega'_{ij}vorticity tensor
τ2ijV\tau^{V}_{2ij}viscous stress tensor
τ2ijE\tau^{E}_{2ij}elastic stress tensor
τ2′\tau'_2stress tensor
αi′\alpha'_i, i=1,…,6i=1,\ldots,6Leslie viscosities (Newtonian: μ′=α4′/2\mu'=\alpha'_4/2, all other αi′=0\alpha'_i=0)
γ1′=α3′−α2′\gamma'_1=\alpha'_3-\alpha'_2rotational/twist viscosity
γt′=α6′−α5′\gamma'_t=\alpha'_6-\alpha'_5torsion coefficient
K1K_1, K3K_3elastic constants for splay and bend respectively

Table A1. Liquid crystal variables and parameters [144].

Boundary conditions

We now list the boundary conditions, proceeding from the cornea out towards the air. At y′=0y'=0, we have velocity continuity and a no flux boundary condition for osmolarity transport. These are, respectively:

u1′=J0(40)\mathbf{u}'_{1}=\mathbf{J}_{0} \tag*{(40)}
c′(u1′−s0t′′)−D1∇′c′=0.(41)c'(\mathbf{u}'_{1}-\mathbf{s}'_{0t'})-D_{1}\nabla'c'=0. \tag*{(41)}

Note that as we assume the cornea is flat, s0t′′=0\mathbf{s}'_{0t'}=0.

At y′=h1′y'=h'_1, we have velocity continuity, aqueous and lipid mass conservation, and the stress balance. In addition, we have a no flux boundary condition for osmolarity transport, and the anchoring boundary condition for the LL. The equations, respectively, are:

(u1′−u2′)⋅t^1′=0,(42)(\mathbf{u}'_1-\mathbf{u}'_2)\cdot\hat{\mathbf{t}}'_1=0, \tag*{(42)}
ρ1(u1′−s1t′′)⋅n^1′=Je′,(43)\rho_1(\mathbf{u}'_1-\mathbf{s}'_{1t'})\cdot\hat{\mathbf{n}}'_1=J'_e, \tag*{(43)}
ρ2(s1t′′−u2′)⋅n^1′=0,(44)\rho_2(\mathbf{s}'_{1t'}-\mathbf{u}'_2)\cdot\hat{\mathbf{n}}'_1=0, \tag*{(44)}
(τ1′−τ2′)⋅n^1′=−γs1′n^1′∇′⋅n^1′+∇s1′γs1′,(45)(\boldsymbol{\tau}'_1-\boldsymbol{\tau}'_2)\cdot\hat{\mathbf{n}}'_1=-\gamma'_{s1}\hat{\mathbf{n}}'_1\nabla'\cdot\hat{\mathbf{n}}'_1+\nabla'_{s1}\gamma'_{s1}, \tag*{(45)}
c′(u1′−s1t′′)−D1∇′c′=0,(46)c'(\mathbf{u}'_1-\mathbf{s}'_{1t'})-D_1\nabla'c'=0, \tag*{(46)}
θ=θB.(47)\theta=\theta_B. \tag*{(47)}

To resolve the stress balance into tangential and normal stress components, we take the dot product of the stress balance and t^1′\hat{\mathbf{t}}'_1 or n^1′\hat{\mathbf{n}}'_1 respectively. Here ∇s1′=(I−n^1′n^1′)⋅∇′\nabla'_{s1}=(\mathbf{I}-\hat{\mathbf{n}}'_1\hat{\mathbf{n}}'_1)\cdot\nabla' is the surface gradient, where I\mathbf{I} is the identity matrix [152].

At y′=h′y'=h', we have lipid mass conservation, stress balance, and anchoring respectively:

ρ2(u2′−s2t′′)⋅n^2′=0,(48)\rho_2(\mathbf{u}'_2-\mathbf{s}'_{2t'})\cdot\hat{\mathbf{n}}'_2=0, \tag*{(48)}
τ2′⋅n^1′=−γ2n^2′∇′⋅n^2′,(49)\boldsymbol{\tau}'_2\cdot\hat{\mathbf{n}}'_1=-\gamma_2\hat{\mathbf{n}}'_2\nabla'\cdot\hat{\mathbf{n}}'_2, \tag*{(49)}
θ=θB.(50)\theta=\theta_B. \tag*{(50)}

The stress balance can be resolved into components in the same way as above, using the unit tangent and normal vectors at this interface, t^2′\hat{\mathbf{t}}'_2 and n^2′\hat{\mathbf{n}}'_2. At the aqueous-lipid interface y′=h1′y'=h'_1, we have surfactant transport and the linear equation of state for the surface tension, respectively:

Γt′+∇s′⋅(Γ′u1′)=Ds∇s′2Γ′(51)\Gamma'_{t}+\nabla'_{s}\cdot(\Gamma'\mathbf{u}'_1)=D_s\nabla_{s}^{\prime2}\Gamma' \tag*{(51)}
γs1′=γ1−RT0Γ′(52)\gamma'_{s1}=\gamma_1-RT_0\Gamma' \tag*{(52)}

Osmosis and evaporation

The effect of osmosis is determined by the difference in concentration on either side of the aqueous-corneal interface y′=0y'=0,

J0′⋅n^0′=Pc(c′−c0).(53)\mathbf{J}'_0\cdot\hat{\mathbf{n}}'_0=P_c(c'-c_0). \tag*{(53)}

Here we assume an ideal osmolarity (c0=300c_0=300 mOsM) on the corneal side of the interface.

To account for the effect of evaporation at the aqueous-lipid interface y′=h1′y'=h'_1, we modify a boundary layer model derived thoroughly in Stapf et al. [135]

Je′=E01+kmDkh2′.(54)J'_e=\frac{E_0}{1+\frac{k_m}{D_k}h'_2}. \tag*{(54)}

Here, DkD_k represents resistance to evaporation provided by the LL based on its thickness and permeability. Its value comes from a nonlinear least squares fit to clinical data [34] performed by Stapf et al. [135], which is very similar to that of Cerretani et al. [123]. We further modify DkD_k to include a dependence on the director angle of liquid crystals in the LL. We suggest the orientation of the molecules affects evaporation through

Je′=E01+kmDk(0.1+0.9sin⁡θB)h2′.(55)J'_e=\frac{E_0}{1+\frac{k_m}{D_k}(0.1+0.9\sin\theta_B)h'_2}. \tag*{(55)}

The maximum evaporation rate occurs when θB=0\theta_B=0, which is ten times that of the rate for θB=π/2\theta_B=\pi/2. The minimum rate matches that used in [135].

Deriving the leading order system

We nondimensionalize using the scalings in (1), along with

W′=Kδ2L2W,αi′=μ2αi,γt′=μ2γt,γr′=μ2γr.W'=\frac{K}{\delta^2L^2}W,\qquad\alpha'_i=\mu_2\alpha_i,\qquad\gamma'_t=\mu_2\gamma_t,\qquad\gamma'_r=\mu_2\gamma_r.

Then, we asymptotically expand all the dependent variables in powers of the small parameter ϵ≪1\epsilon\ll1. For example,

Γ(x,t)∼Γ0(x,t)+ϵΓ1(x,t)+ϵ2Γ2(x,t)+⋯\Gamma(x,t)\sim\Gamma_0(x,t)+\epsilon\Gamma_1(x,t)+\epsilon^2\Gamma_2(x,t)+\cdots

We substitute these expansions into the nondimensionalized equations and collect like powers of ϵ\epsilon. At leading order, we are able to determine the director angle, conservation of mass in both the AL and LL, and conservation of surfactant concentration. At O(ϵ)O(\epsilon), we are only able to determine intermediate variables, so we proceed to O(ϵ2)O(\epsilon^{2}). We are then able to determine the AL depth averaged velocity, AL pressure, the force balance equation for the axial velocity in the LL, and conservation of osmolarity. This gives us the closed system of equations from § 2.1. This progression is typical of problems like this [100, 135].

References

  1. [1]Bron, A.J., Tiffany, J.M., Gouveia, S.M., Yokoi, N., Voon, L.W.: Functional aspects of the tear film lipid layer. Exp. Eye Res. 78, 347–360 (2004)
  2. [2]Smith, J.A., et al.: The epidemiology of dry eye disease: Report of the Epidemiology Subcommittee of the International Dry Eye WorkShop. Ocul. Surf. 5, 93–107 (2007)
  3. [3]Stapleton, F., Alves, M., Bunya, V.Y., Jalbert, I., Lekhanont, K., Malet, F., Na, K.-S., Schaumberg, D., Uchino, M., Vehof, J., Viso, E., Vitale, S., Jones, L.: TFOS DEWS-II epidemiology report. Ocul. Surf. 15(3), 334–365 (2017)DOI
  4. [4]Lemp, M.A.: Report of the National Eye Institute/Industry workshop on Clinical Trials in Dry Eyes. CLAO J 21, 221–232 (1995)
  5. [5]Lemp, M.A., et al.: The definition and classification of dry eye disease: Report of the Definition and Classification Subcommittee of the International Dry Eye WorkShop. Ocul. Surf. 5, 75–92 (2007)
  6. [6]Craig, J.P., Nichols, K.K., Akpek, E.K., Caffery, B., Dua, H.S., Joo, C.-K., Liu, Z., Nelson, J.D., Nichols, J.J., Tsubota, K., et al.: TFOS DEWS-II definition and classification report. Ocul. Surf. 15(3), 276–283 (2017)DOI
  7. [7]Willcox, M.D.P., Argüeso, P., Georgiev, G.A., Holopainen, J.M., Laurie, G.W., Millar, T.J., Papas, E.B., Rolland, J.P., Schmidt, T.A., Stahl, U., Suarez, T., Subbaraman, L.N., Ucakhan, O.O., Jones, L.W.: TFOS DEWS-II Tear Film Report. Ocul. Surf. 15(3), 369–406 (2017)DOI
  8. [8]Baudouin, C., Aragona, P., Messmer, E.M., Tomlinson, A., Calonge, M., Boboridis, K.G., Akova, Y.A., Geerling, G., Labetoulle, M., Rolando, M.: Role of hyperosmolarity in the pathogenesis and management of dry eye disease: Proceedings of the OCEAN group meeting. Ocul. Surf. 11, 246–258 (2013)DOI
  9. [9]Tsubota, K.: Tear dynamics and dry eye. Prog. Retin. Eye Res. 17(4), 565–596 (1998)
  10. [10]Braun, R.J., King-Smith, P.E., Begley, C.G., Li, L., Gewecke, N.R.: Dynamics and function of the tear film in relation to the blink cycle. Prog. Retin. Eye Res. 45, 132–164 (2015)
  11. [11]Mishima, S., Maurice, D.M.: The oily layer of the tear film and evaporation from the corneal surface. Exp. Eye Res. 1, 39–45 (1961)
  12. [12]Svitova, T.F., Lin, M.C.: Evaporation retardation by model tear-lipid films: The roles of film aging, compositions and interfacial rheological properties. Coll. Surf. B: Biointerfaces 197, 111392 (2021)DOI
  13. [13]Norn, M.S.: Semiquantitative interference study of the fatty layer of precorneal film. Acta Ophthalmol. 57, 766–774 (1979)
  14. [14]Berger, R.E., Corrsin, S.: A surface tension gradient mechanism for driving the pre-corneal tear film after a blink. J. Biomech. 7, 225–238 (1974)
  15. [15]Craster, R.V., Matar, O.K.: Dynamics and stability of thin liquid films. Rev. Mod. Phys. 81(3), 1131 (2009)
  16. [16]Holly, F.J.: Formation and rupture of the tear film. Exp. Eye Res. 15, 515–525 (1973)
  17. [17]Tiffany, J.M.: Measurement of wettability of the corneal epithelium: I. particle attachment method. Acta Ophthalmol. 68(2), 175–181 (1990)
  18. [18]Tiffany, J.M.: Measurement of wettability of the corneal epithelium: II. contact angle method. Acta Ophthalmol. 68(2), 182–187 (1990)
  19. [19]Gipson, I.K.: Distribution of mucins at the ocular surface. Exp. Eye Res. 78, 379–388 (2004)
  20. [20]King-Smith, P.E., Fink, B.A., Hill, R.M., Koelling, K.W., Tiffany, J.M.: The thickness of the tear film. Curr. Eye. Res. 29, 357–368 (2004)
  21. [21]Yokoi, N., Takehisa, Y., Kinoshita, S.: Correlation of tear lipid layer interference patterns with the diagnosis and severity of dry eye. Am. J. Ophthalmol. 122, 818–824 (1996)
  22. [22]Goto, E., Tseng, S.C.G.: Kinetic analysis of tear interference images in aqueous tear deficiency dry eye before and after punctal occlusion. Invest. Ophthalmol. Vis. Sci. 44, 1897–1905 (2003)
  23. [23]King-Smith, P.E., Nichols, J.J., Nichols, K.K., Braun, R.J.: A high resolution microscope for imaging the lipid layer of the tear film. Ocul. Surf. 9(4), 197–211 (2011)
  24. [24]Govindarajan, B., Gipson, I.K.: Membrane-tethered mucins have multiple functions on the ocular surface. Exp. Eye Res. 90, 655–693 (2010)DOI
  25. [25]Owens, H., Phillips, J.: Spreading of the tears after a blink: Velocity and stabilization time in healthy eyes. Cornea 20, 484–487 (2001)
  26. [26]King-Smith, P.E., Nichols, J.J., Nichols, K.K., Fink, B.A., Braun, R.J.: Contributions of evaporation and other mechanisms to tear film thinning and breakup. Optom. Vis. Sci. 85, 623–630 (2008)
  27. [27]Norn, M.S.: Dessication of the precorneal film. I Corneal wetting-time. Acta Ophthalmol. 47, 865–80 (1969)DOI
  28. [28]Lemp, M.A., et al.: The definition and classification of dry eye disease: Report of the Definition and Classification Subcommittee of the International Dry Eye WorkShop. Ocul. Surf. 5, 75–92 (2007)
  29. [29]King-Smith, P.E., Begley, C.G., Braun, R.J.: Mechanisms, imaging and structure of tear film breakup. Ocul. Surf. 16, 4–30 (2018)
  30. [30]Zhong, L., Braun, R.J., Begley, C.G., King-Smith, P.E.: Dynamics of fluorescent imaging for rapid tear thinning. Bull. Math. Biol. 81, 39–80 (2019)DOI
  31. [31]Zhang, Y.L., Matar, O.K., , Craster, R.V.: Analysis of tear film rupture: Effect of non-newtonian rheology. J. Colloid Interface Sci. 262, 130–148 (2003)
  32. [32]Zhang, Y.L., Matar, O.K., , Craster, R.V.: Rupture analysis of the corneal mucus layer of the tear film. Mol. Simul. 30, 167–172 (2004)DOI
  33. [33]Yokoi, N., Georgiev, G.A.: Tear-film-oriented diagnosis for dry eye. Japan. J. Ophthalmol. 63, 127–136 (2019)
  34. [34]King-Smith, P.E., Hinel, E.A., Nichols, J.J.: Application of a novel interferometric method to investigate the relation between lipid layer thickness and tear film thinning. Invest. Ophthalmol. Vis. Sci. 51, 2418–23 (2010)DOI
  35. [35]Sharma, A., Ruckenstein, E.: Mechanism of tear film rupture and formation of dry spots on cornea. J. Coll. Interface Sci. 106, 12–27 (1985)DOI
  36. [36]Sharma, A., Khanna, R., Reiter, G.: A thin film analog of the corneal mucus layer of the tear film: an enigmatic long range non-classical DLVO interaction in the breakup of thin polymer films. Coll. Surf. B 14, 223–35 (1999)DOI
  37. [37]Begley, C.G., Simpson, T., Liu, H., Salvo, E., Wu, Z., Bradley, A., Situ, P.: Quantitative Analysis of Tear Film Fluorescence and Discomfort During Tear Film Instability and Thinning. Invest. Ophthalmol. Vis. Sci. 54(4), 2645–2653 (2013)DOI
  38. [38]King-Smith, P.E., Bailey, M.D., Braun, R.J.: Four characteristics and a model of an effective tear film lipid layer. Ocul. Surf. 11, 1–10 (2013)
  39. [39]Luke, R.A., Braun, R.J., Driscoll, T.A., Awisi-Gyau, D., Begley, C.G.: Parameter Estimation for Mixed-Mechanism Tear Film Thinning. Bull. Math. Biol. 83(5), 56 (2021)DOI
  40. [40]Doane, M.G.: Blinking and the mechanics of the lacrimal drainage system. Ophthalmol. 88, 844–851 (1981)
  41. [41]Tomlinson, A., Doane, M.G., McFayden, A.: Inputs and outputs of the lacrimal system: Review of production and evaporative loss. Ocul. Surf. 7, 17–29 (2009)
  42. [42]Kimball, S.H., King-Smith, P.E., Nichols, J.J.: Evidence for the major contribution of evaporation to tear film thinning between blinks. Invest. Ophthalmol. Vis. Sci. 51, 6294–97 (2010)DOI
  43. [43]Braun, R.J.: Dynamics of the tear film. Annu. Rev. Fluid Mech. 44, 267–297 (2012)
  44. [44]Cerretani, C.F., Radke, C.J.: Tear dynamics in healthy and dry eyes. Curr. Eye Res. 39, 580–595 (2014)
  45. [45]Stahl, U., Willcox, M., Stapleton, F.: Osmolality and tear film dynamics. Clin. Exp. Optom. 95, 3–11 (2012)
  46. [46]Peng, C.-C., Cerretani, C., Braun, R.J., Radke, C.J.: Evaporation-driven instability of the precorneal tear film. Adv. Colloid Interface Sci. 206, 250–264 (2014)DOI
  47. [47]Lemp, M.A., Bron, A.J., Baudouin, C., Castillo, J.M.B., Geffen, D., Tauber, J., Foulks, G.N., Pepose, J.S., Sullivan, B.D.: Tear osmolarity in the diagnosis and management of dry eye disease. Am. J. Ophthalmol. 151(5), 792–798 (2011)DOI
  48. [48]Tomlinson, A., Khanal, S., Ramaesh, K., Diaper, C., McFayden, A.: Tear film osmolarity: determination of a referent for dry eye diagnosis. Invest. Ophthalmol. Vis. Sci. 47(10), 4309–4315 (2006)
  49. [49]Versura, P., Profazio, V., Campos, E.: Performance of tear osmolarity compared to previous diagnostic tests for dry eye diseases. Curr. Eye Res. 35(7), 553–564 (2010)DOI
  50. [50]Gilbard, J.P., Farris, R.L., Santamaria, J.: Osmolarity of tear microvolumes in keratoconjunctivitis sicca. Arch. Ophthalmol. 96, 677–681 (1978)
  51. [51]Sullivan, B.D., Whitmer, D., Nichols, K.K., Tomlinson, A., Foulks, G.N., Geerling, G., Pepose, J.S., Kosheleff, V., Porreco, A., Lemp, M.A.: An objective approach to dry eye disease severity. Invest. Ophthalmol. Vis. Sci. 51(12), 6125–6130 (2010)DOI
  52. [52]Dartt, D., Willcox, M.: Complexity of the tear film: importance in homeostasis and dysfunction during disease. Exp. Eye Res. 117, 1–3 (2013)
  53. [53]Liu, H., Begley, C., Chen, M., Bradley, A., Bonanno, J., McNamara, N.A., Nelson, J.D., Simpson, T.: A link between tear instability and hyperosmolarity in dry eye. Invest. Ophthalmol. Vis. Sci. 50(8), 3671–3679 (2009)DOI
  54. [54]Luke, R.A., Braun, R.J., Driscoll, T.A., Begley, C.G., Awisi-Gyau, D.: Parameter Estimation for Evaporation-Driven Tear Film Thinning. Bull. Math. Biol. 82(6), 71 (2020)DOI
  55. [55]Pflugfelder, S.C.: Tear dysfunction and the cornea: LXVIII Edward Jackson memorial lecture. Am. J. Ophthalmol. 152(6), 900–909 (2011)DOI
  56. [56]Belmonte, C., Acosta, M.C., Merayo-Lloves, J., Gallar, J.: What causes eye pain? Curr. Ophthalmol. Rep. 3(2), 111–121 (2015)
  57. [57]Nichols, J.J., Mitchell, G.L., King-Smith, P.E.: Thinning rate of the precorneal and prelens tear films. Invest. Ophthalmol. Vis. Sci. 46(7), 2353–2361 (2005)
  58. [58]Hamano, H., Hori, M., Mitsunaga, S.: Measurement of evaporation rate of water from the precorneal tear film and contact lenses. Contacto 25(2), 7–15 (1981)
  59. [59]Peng, C.-C., Cerretani, C., Li, Y., Bowers, S., Shahsavarani, S., Lin, M., Radke, C.: Flow evaporimeter to assess evaporative resistance of human tear-film lipid layer. Ind. Eng. Chem. Res. 53(47), 18130–18139 (2014)DOI
  60. [60]Dursch, T.J., Li, W., Taraz, B., Lin, M.C., Radke, C.J.: Tear-film evaporation rate from simultaneous ocular-surface temperature and tear-breakup area. Optom. Vis. Sci. 95, 5–12 (2018)DOI
  61. [61]Braun, R.J., Driscoll, T.A., Begley, C.G., King-Smith, P.E., Siddique, J.I.: On tear film breakup (TBU): Dynamics and imaging. Math. Med. Biol. 45, 145–180 (2018)
  62. [62]Tiffany, J.M., Dart, J.K.G.: Normal and abnormal functions of meibomian secretion. R. Soc. Med. Int. Congr. Symp. 40, 1061–4 (1981)
  63. [63]Purslow, C., Wolffsohn, J.S.: Ocular surface temperature: A review. Eye Contact Lens 31, 117–123 (2005)
  64. [64]Tiffany, J.M.: Viscoelastic properties of human tears and polymer solutions. In: Sullivan, D.A. (ed.) Lacrimal Gland, Tear Film, and Dry Eye Syndromes, pp. 267–70. Plenum Press, New York (1994)
  65. [65]Pandit, J.C., Nagyová, B., Bron, A.J., Tiffany, J.M.: Physical properties of stimulated and unstimulated tears. Exp. Eye Res. 68, 247–53 (1999)
  66. [66]Tiffany, J.M.: The viscosity of human tears. Intl. Ophthalmol. 15, 371–376 (1991)
  67. [67]Jossic, L., Lefevre, P., Loubens, C., Magnin, A., Corre, C.: The fluid mechanics of shear-thinning tear substitutes. J. Nonnewton. Fluid Mech. 161, 1–9 (2009)DOI
  68. [68]Tiffany, J.M., Nagyová, B.: Components of tears responsible for surface tension. In: Sullivan, D.A., Dartt, D.A., Meneray, M.A. (eds.) Lacrimal Gland, Tear Film, and Dry Eye Syndromes 2, pp. 581–5. Plenum Press, New York (1998)
  69. [69]Tiffany, J.M.: The lipid secretion of the meibomian glands. Adv. Lipid Res. 22, 1–62 (1987)
  70. [70]Leiske, D.L., Leiske, C.I., Leiske, D.R., Toney, M.F., Senchyna, M., Ketelson, H.A., Meadows, D.L., Fuller, G.G.: Temperature-induced transitions in the structure and interfacial rheology of human meibum. Biophys. J. 102, 369–376 (2011)
  71. [71]Leiske, D.L., Miller, C.E., Rosenfeld, L., Cerretani, C., Ayzner, A., Lin, B., Meron, M., Senchyna, M., Ketelson, H.A., Meadows, D., Srinivasan, S., Jones, L., Radke, C.J., Toney, M.F., Fuller, G.G.: Molecular structure of interfacial human meibum films. Langmuir 28, 11858–11865 (2012)
  72. [72]Rosenfeld, L., Cerretani, C., Leiske, D.L., Toney, M.F., Radke, C.J., Fuller, G.G.: Structural and rheological properties of meibomian lipid. Invest. Ophthalmol. Vis. Sci. 54(4), 2720–2732 (2013)
  73. [73]Miller, D.: Measurement of the surface tension of tears. Arch. Ophthalmol. 82, 368–71 (1969)
  74. [74]Nagyová, B., Tiffany, J.M.: Components of tears responsible for surface tension. Curr. Eye Res. 19, 4–11 (1999)
  75. [75]Mudgil, P., Millar, T.J.: Adsorption of lysozyme to phospholipid and Meibomian lipid monolayer films. Coll. Surf. B: Biointerfaces 48, 128–37 (2006)
  76. [76]Mudgil, P., Millar, T.J.: The effect of Meibomian lipids and tear proteins on evaporation rate under controlled in vitro conditions. Exp. Eye Res. 86, 622–8 (2008)
  77. [77]Butovich, I.A.: Tear film lipids. Experimental Eye Research 117, 4–27 (2013)
  78. [78]Rantamäki, A.H., Holopainen, J.M.: The Effect of Phospholipids on Tear Film Lipid Layer Surface Activity. Invest. Ophthalmol. Vis. Sci. 58, 149–154 (2017)
  79. [79]Galor, A., Sanchez, V., Jensen, A., Burton, M., Maus, K., Stephenson, D., Chalfant, C., Mandal, N.: Meibum sphingolipid composition is altered in individuals with meibomian gland dysfunction-a side by side comparison of meibum and tear sphingolipids. The Ocular Surface 23, 87–95 (2022)
  80. [80]Georgiev, G.A., Eftimov, P., Yokoi, N.: Structure-function relationship of tear film lipid layer: A contemporary perspective. Exp. Eye Res. 163, 17–28 (2017)DOI
  81. [81]Braun, R.J., Driscoll, T.A., Begley, C.G.: Mathematical models of the tear film. In: Ocular Fluid Dynamics, pp. 387–432. Springer, Berlin (2019)
  82. [82]Braun, R.J., Usha, R., McFadden, G.B., Driscoll, T.A., Cook, L.P., King-Smith, P.E.: Thin film dynamics on a prolate spheroid with application to the cornea. J. Eng. Math. 73, 121–138 (2012)DOI
  83. [83]Maki, K.L., Braun, R.J., Henshaw, W.D., King-Smith, P.E.: Tear film dynamics on an eye-shaped domain I: pressure boundary conditions. Math. Med. Biol. 27, 227–254 (2010)DOI
  84. [84]Maki, K.L., Braun, R.J., Ucciferro, P., Henshaw, W.D., King-Smith, P.E.: Tear film dynamics on an eye-shaped domain. Part 2. Flux boundary conditions. J. Fluid Mech. 647, 361–390 (2010)
  85. [85]Li, L., Braun, R.J., Maki, K.L., Henshaw, W.D., King-Smith, P.E.: Tear film dynamics with evaporation, wetting and time-dependent flux boundary condition on an eye-shaped domain. Phys. Fluids 26, 052101 (2014)
  86. [86]Li, L., Braun, R.J., Henshaw, W.D., King-Smith, P.E.: Computed tear film and osmolarity dynamics on an eye-shaped domain. Math. Med. Biol. 33, 123–157 (2016)DOI
  87. [87]Li, L., Braun, R.J., Henshaw, W.D., King-Smith, P.E.: Computed flow and fluorescence over the ocular surface. Math. Med. Biol. 35, 51–85 (2018)
  88. [88]Driscoll, T.A., Braun, R.J., Brosch, J.K.: Simulation of parabolic flow on an eye-shaped domain with moving boundary. J. Eng. Math. 111(1), 111–126 (2018)
  89. [89]Maki, K.L., Henshaw, W., McManus, A., Braun, R., Chapp, D., Driscoll, T.: A model for tear film dynamics during a realistic blink. Modeling and Artificial Intelligence in Ophthalmology 2(3), 21–27 (2019)
  90. [90]Gaffney, E.A., Tiffany, J.M., Yokoi, N., Bron, A.J.: A mass and solute balance model for tear volume and osmolarity in the normal and the dry eye. Prog. Retinal Eye Res. 29, 59–78 (2010)
  91. [91]Sharma, A., Tiwari, S., Khanna, R., Tiffany, J.M.: Hydrodynamics of Meniscus-induced Thinning of the Tear Film. In: Sullivan, D.A., Dartt, D.A., Meneray, M.A. (eds.) Advances in Experimental Medicine and Biology. Lacrimal Gland, Tear Film, and Dry Eye Syndromes 2, vol. 438, pp. 425–431. Springer, Berlin (1998)
  92. [92]Wong, H., Fatt, I., Radke, C.: Deposition and thinning of the human tear film. J. Colloid Interface Sci. 184(1), 44–51 (1996)
  93. [93]Miller, K.L., Polse, K.A., Radke, C.J.: Black line formation and the “perched” human tear film. Curr. Eye Res. 25, 155–62 (2002)
  94. [94]Braun, R.J., Fitt, A.D.: Modeling the drainage of the precorneal tear film after a blink. Math. Med. Biol. 20, 1–28 (2003)
  95. [95]Li, L., Braun, R.J.: A model for the human tear film with heating from within the eye. Phys. Fluids 24, 062103 (2012)DOI
  96. [96]Jones, M.B., Please, C.P., McElwain, D.L.S., Fulford, G.R., Roberts, A.P., Collins, M.J.: Dynamics of tear film deposition and draining. Math. Med. Biol. 22, 265–88 (2005)
  97. [97]Jones, M.B., McElwain, D.L.S., Fulford, G.R., Collins, M.J., Roberts, A.P.: The effect of the lipid layer on tear film behavior. Bull. Math. Biol. 68, 1355–81 (2006)
  98. [98]Maki, K.L., Braun, R.J., Driscoll, T.A., King-Smith, P.E.: An overset grid method for the study of reflex tearing. Math. Med. Biol. 25, 187–214 (2008)DOI
  99. [99]Aydemir, E., Breward, C.J.W., Witelski, T.P.: The effect of polar lipids on tear film dynamics. Bull. Math. Biol. 73, 1171–1201 (2010)
  100. [100]Bruna, M., Breward, C.J.W.: The influence of non-polar lipids on tear film dynamics. J. Fluid Mech. 746, 565–605 (2014)
  101. [101]Allouche, M., Abderrahmane, H.A., Djouadi, S.M., Mansouri, K.: Influence of curvature on tear film dynamics. Eur. J. Mech. B Fluids 66, 81–91 (2017)
  102. [102]Mehdaoui, H., Abderrahmane, H.A., Bouda, F.N., Koulali, A.: 2D numerical simulation of tear film dynamics: effects of shear-thinning properties. Eur. J. Mech. B Fluids 90, 128–136 (2021)
  103. [103]Braun, R.J., King-Smith, P.E.: Model problems for the tear film in a blink cycle: Single equation models. J. Fluid Mech. 586, 465–90 (2007)
  104. [104]Heryudono, A., Braun, R.J., Driscoll, T.A., Maki, K.L., Cook, L.P., King-Smith, P.E.: Single-equation models for the tear film in a blink cycle: realistic lid motion. Math. Med. Biol. J. IMA 24(4), 347–377 (2007)
  105. [105]Zubkov, V.S., Breward, C.J.W., Gaffney, E.A.: Coupling fluid and solute dynamics within the ocular surface tear film: a modelling study of black line osmolarity. Bull. Math. Biol. 74(9), 2062–2093 (2012)
  106. [106]Deng, Q., Braun, R.J., Driscoll, T.A.: Heat transfer and tear film dynamics over multiple blink cycles. Phys. Fluids 26(7), 071901 (2014)DOI
  107. [107]Deng, Q., Braun, R.J., Driscoll, T.A., King-Smith, P.E.: A model for the tear film and ocular surface temperature for partial blinks. Interfacial Phenom. Heat Transf. 1(4) (2013)
  108. [108]Luke, R.A., Braun, R.J., Begley, C.G.: Mechanistic determination of tear film thinning via fitting simplified models to tear breakup. Model. Artif. Int. Ophthalmol. 3, 71–100 (2021)
  109. [109]Driscoll, T.A., Braun, R.J., Luke, R.A., Sinopoli, D., Phatak, A., Dorsch, J., Begley, C.G., Awisi-Gyau, D.: Fitting ODE models of tear film breakup. Model. Art. Int. Ophthal. 5(1), 1–36 (2023)
  110. [110]Zhong, L., Ketelaar, C.F., Braun, R.J., Begley, C.G., King-Smith, P.E.: Mathematical modeling of glob-driven tear film breakup. Math. Med. Biol., 1–37 (2018)DOI
  111. [111]Zhong, L., Braun, R.J., King-Smith, P.E., Begley, C.G.: Mathematical modeling of glob-driven tear film breakup. J. Model. Ophthalmol. 2(1), 24–28 (2018)
  112. [112]Dey, M., Vivek, A.S., Dixit, H.N., Richhariya, A., Feng, J.J.: A model of tear-film breakup with continuous mucin concentration and viscosity profiles. J. Fluid Mech. 858, 352–376 (2019)
  113. [113]Dey, M., Vivek, A.S., Dixit, H.N., Richhariya, A., Feng, J.J.: A model of tear-film breakup with continuous mucin concentration and viscosity profiles–Corrigendum. J. Fluid Mech. 889, 1 (2020)
  114. [114]Choudhury, A., Dey, M., Dixit, H.N., Feng, J.J.: Tear-film breakup: The role of membrane-associated mucin polymers. Phys. Rev. E 103, 013018 (2021)
  115. [115]Ji, H., Witelski, T.P.: Finite-time thin film rupture driven by modified evaporative loss. Physica D: Nonlinear Phenomena 342, 1–15 (2017)
  116. [116]Ji, H., Witelski, T.P.: Steady states and dynamics of a thin-film-type equation with non-conserved mass. Eur. J. Appl. Math. 31, 968–1001 (2020)
  117. [117]Shi, X., Rodríguez-Hakim, M., Shaqfeh, E.S.G., Fuller, G.G.: Instability and symmetry breaking in binary evaporating thin films over a solid spherical dome. J. Fluid Mech. 915, 45 (2021)
  118. [118]Shi, X., Fuller, G.G., Shaqfeh, E.S.G.: Instability and symmetry breaking of surfactant films over an air bubble. J. Fluid Mech. 953, 26 (2022)
  119. [119]Pucker, A., Haworth, K.: The presence and significance of polar meibum and tear lipids. Ocul. Surf. 13(1), 26–42 (2015)
  120. [120]Butovich, I.A.: Meibomian glands, meibum, and meibogenesis. Exp. Eye Res. 163, 2–16 (2017)
  121. [121]Georgiev, G.A., Borchman, D., Eftimov, P., Yokoi, N.: Lipid saturation and the rheology of human tear lipids. Int. J. Molec. Sci. 20, 3431 (2019)
  122. [122]McCulley, J.P., Shine, W.: A compositional based model for the tear film lipid layer. Tr. Am. Ophthalmol. Soc. XCV, 79–93 (1997)
  123. [123]Cerretani, C.F., Ho, N.H., Radke, C.: Water-evaporation reduction by duplex films: Application to the human tear film. Adv. Colloid Interface Sci. 197, 33–57 (2013)
  124. [124]Sledge, S.M., Khimji, H., Borchman, D., Oliver, A.L., Michael, H., Dennis, E., Gerlach, D., Bhola, R., Stephen, E.: Evaporation and hydrocarbon chain conformation of surface lipid films. Ocul. Surf. 14(4), 447–459 (2016)
  125. [125]Xu, X., Li, G., Zuo, Y.Y.: Effect of Model Tear Film Lipid Layer on Water Evaporation. Invest. Ophthalmol. Vis. Sci. 64(1), 13 (2023)
  126. [126]King-Smith, P.E., Reuter, K.S., Braun, R.J., Nichols, J.J., Nichols, K.K.: Tear film breakup and structure studied by simultaneous video recording of fluorescence and tear film lipid layer, TFLL, images. Invest. Ophthalmol. Vis. Sci. 54, 4900–4909 (2013)
  127. [127]Fenner, B.J., Tong, L.: More to Stable Tears Than Thickness of the Tear Film Lipid Layer. Invest. Ophthalmol. Vis. Sci 56(3), 1601–1601 (2015)
  128. [128]King-Smith, P.E., Braun, R.J.: Author Response: More to Stable Tears Than Thickness of the Lipid Layer. Invest. Ophthalmol. Vis. Sci 56(3), 1602–1602 (2015)
  129. [129]Wizert, A., Iskander, D., Cwiklik, L.: Organization of lipids in the tear film: A molecular-level view. PLoS ONE 9, 92461 (2014)
  130. [130]Cwiklik, L.: Tear film lipid layer: A molecular level view. Biochimica et Biophysica Acta 1858, 2421–2430 (2016)
  131. [131]Bland, H.C., Moilanen, J.A., Ekholm, F.S., Paananen, R.O.: Investigating the role of specific tear film lipids connected to dry eye syndrome: A study on o-acyl-ω-hydroxy fatty acids and diesters. Langmuir 35(9), 3545–3552 (2019)
  132. [132]Viitaja, T., Moilanen, J., Svedström, K.J., Ekholm, F.S., Paananen, R.O.: Tear film lipid layer structure: Self-assembly of o-acyl-ω-hydroxy fatty acids and wax esters into evaporation-resistant monolayers. Nano Letters 21(18), 7676–7683 (2021)
  133. [133]Paananen, R.O., Javanainen, M., Holopainen, J.M., Vattulainen, I.: Crystalline wax esters regulate the evaporation resistance of tear film lipid layers associated with dry eye syndrome. J. Phys. Chem. Lett. 10(14), 3893–3898 (2019)
  134. [134]Butovich, I.A., Lu, H., McMahon, A., Ketelson, H., Senchyna, M., Meadows, D., Campbell, E., Molai, M., Linsenbart, E.: Biophysical and morphological evaluation of human normal and dry eye meibum using hot stage polarized light microscopy. Invest. Ophthalmol. Vis. Sci. 55(1), 87–101 (2014)
  135. [135]Stapf, M.R., Braun, R.J., King-Smith, P.E.: Duplex tear film evaporation analysis. Bull. Math. Biol. 79(12), 2814–2846 (2017)
  136. [136]Taranchuk, M.J., Braun, R.J.: On the effect of liquid crystal orientation in the lipid layer on tear film thinning and breakup. ArXiv (2404.13225 [physics.flu-dyn]) (2024)
  137. [137]Matar, O.K., Craster, R.V., Warner, M.R.E.: Surfactant transport on highly viscous surface films. J. Fluid Mech. 466, 85–111 (2002)
  138. [138]Cummings, L.J., Low, J., Myers, T.G.: Extensional flow of nematic liquid crystal with an applied electric field. Eur. J. Appl. Math. 25(4), 397–423 (2014)
  139. [139]Taranchuk, M.J., Cummings, L.J., Driscoll, T.A., Braun, R.J.: Extensional flow of a free film of nematic liquid crystal with moderate elasticity. Phys. Fluids 35(6) (2023)
  140. [140]King-Smith, P.E., Hinel, E.A., Nichols, J.J.: Application of a novel interferometric method to investigate the relation between lipid layer thickness and tear film thinning. Invest. Ophthalmol. Vis. Sci. 51, 2418–2423 (2010)
  141. [141]King-Smith, P.E., Nichols, J.J., Braun, R.J., Nichols, K.K.: High resolution microscopy of the lipid layer of the tear film. Ocul. Surf. 9(4), 197–211 (2011)
  142. [142]Goto, E., Tseng, S.C.G.: Kinetic analysis of tear interference images in aqueous tear deficiency dry eye before and after punctal occlusion. Invest. Ophthalmol. Vis. Sci. 44, 1897–1905 (2003)
  143. [143]King-Smith, E., Fink, B., Hill, R., Koelling, K., Tiffany, J.: The thickness of the tear film. Curr. Eye Res. 29(4-5), 357–368 (2004)
  144. [144]Stewart, I.W.: The Static and Dynamic Continuum Theory of Liquid Crystals: a Mathematical Introduction. CRC Press, Boca Raton, FL (2019)
  145. [145]Weast, R.C., Astle, M.J.: CRC Handbook of Chemistry and Physics: A Ready-reference Book of Chemical and Physical Data. CRC Press, Boca Raton, FL (1978)
  146. [146]Tiffany, J.M.: The lipid secretion of the meibomian glands. Adv. Lipid Res. 22, 1–62 (1987)
  147. [147]Nagyova, B., Tiffany, J.M.: Components responsible for the surface tension of human tears. Curr. Eye Res. 19(1), 4–11 (1999)
  148. [148]Riquelme, R., Lira, I., Pérez-López, C., Rayas, J.A., Rodríguez-Vera, R.: Interferometric measurement of a diffusion coefficient: comparison of two methods and uncertainty analysis. J. Phys. D: Appl. Phys. 40(9), 2769 (2007)
  149. [149]Braun, R.J., Naire, S., Snow, S.A.: Limiting cases of gravitational drainage of a vertical free film for evaluating surfactants. SIAM J. Appl. Math. 61, 889–913 (2000)
  150. [150]Stebe, K.J., Maldarelli, C.: Remobilizing surfactant retarded fluid particle interfaces: II. controlling the surface mobility at interfaces of solutions containing surface active components. J. Coll. Interface Sci. 163(1), 177–189 (1994)
  151. [151]Paananen, R.O., Viitaja, T., Olyzynska, A., Ekholm, F.S., Moilanen, J., Cwiklik, L.: Interactions of polar lipids with cholesteryl ester multilayers elucidate tear film lipid layer structure. Ocul. Surf. 18, 545–553 (2020)
  152. [152]Stone, H.: A simple derivation of the time-dependent convective-diffusion equation for surfactant transport along a deforming interface. Phys. Fluids A 2, 111–112 (1990)

Paper details

vlnc:W_h4cd5ap6
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents