Influence of dispersion medium structure on the physicochemical properties of aging colloidal suspensions investigated using the synthetic clay Laponite^®

Laponite XLG^®powder (purchased from BYK Additives Inc., molecular weight- 2286.9 g/mol) was dried in an oven at $120$^∘C for 18 h to remove moisture. The dried powder was weighed and added to Milli-Q water (Millipore Corp., resistivity 18.2 M$\Omega$-cm) which was continually agitated using a magnetic stirrer for 45 min. The additives glucose (Sigma-Aldrich), DMF (SDFCL Fine Chem Pvt.Ltd), NaCl (LABORT Fine Chem Pvt.Ltd.), and KCl (Sigma-Aldrich) were measured and added to de-ionized water prior to the addition of Laponite powder. All the additives were used as received without any further purification. Experiments at different suspensions temperatures were performed following two different protocols. In the first protocol, the temperature of the water was set before the addition of Laponite powder and was maintained throughout the stirring time (45 min). We label these as pre-heated suspensions. In the second protocol, the temperature of the suspension was set after stirring of the sample was stopped. These are labelled as the post-heated suspensions.

Dynamic light scattering experiments were performed using a Brookhaven Instruments Corporation (BIC) setup whose details are given elsewhere D_Saha_2014 ; D_Saha_2015 . A glass cuvette filled with the sample was held in a refractive index matching bath filled with decaline. For the DLS experiments, a freshly prepared sample was filtered into the glass cuvette through a Millipore filter of diameter 0.45 $\mu$m using a syringe. The increase in the sample waiting time, $t_{w}$, was monitored continuously from the time at which filtration was stopped ($t_{w}$ = 0 at the completion of the filtration procedure). The cuvette was sealed and quickly placed in the DLS instrument to record the decays of the intensity autocorrelation functions with increasing $t_{w}$. The temperature of the sample was controlled with a temperature controller equipped with a water circulation unit. A digital autocorrelator was used to measure the intensity autocorrelation function of the scattered light: $g^{(2)}(q,t)=\frac{<I(q,0)I(q,t)>}{<I(q,0)>^{2}}=1+A|g^{(1)}(q,t)|^{2}$ B. J. Berne_1975 . Here $q$, $I(q,t)$, $g^{(1)}(q,t)$ and $A$ are the scattering wave vector, the intensity at a delay time $t$, the normalized electric field autocorrelation function and the coherence factor, respectively. The scattering wave vector $q$ is related to the scattering angle $\theta$, $q=(4\pi n/\lambda)\sin(\theta/2)$, where $n$ and $\lambda$ are the refractive index of the medium and the wavelength of the laser, respectively. The decays of the normalized intensity autocorrelation functions, $C(t)=\frac{g^{(2)}(q,t)-1}{A}$, were measured for Laponite suspensions at different $t_{w}$. The normalized autocorrelation functions for 12.2 mM (2.8% w/v) aqueous Laponite suspensions at several $t_{w}$ are plotted $vs.$ delay time $t$ in Figure 1.

The normalized autocorrelation functions, $C(t)$ plotted in Figure 1, fit best to functions representing two-step decays for the entire range of waiting times explored. This indicates the presence of two relaxation time-scales corresponding to two distinct dynamical processes which can be fitted to the following equation D_Saha_2014 ; D_Saha_2015 ; P_Gadige_2017

Fits to the above equation were used to extract $\tau_{1}$ and $\tau_{ww}$ , identified as the time-scales associated with the $\beta$- and $\alpha$- relaxation processes respectively. The faster $\beta$- relaxation involves the rattling motion of each particle inside its cage. The slower $\alpha$- relaxation corresponds to the time-scale at which the particle diffuses out of its cage in a process facilitated by the cooperative rearrangements of the surrounding particles. The parameters $a$ and $(1-a)$ represent, respectively, the relative strengths of $\beta$- and $\alpha$- relaxation processes, while $\beta$ in eq 1 is a stretching exponent that quantifies the distribution of the $\alpha$- relaxation time-scales. The average $\alpha$- relaxation time can be defined as $<\tau_{ww}>=(\frac{\tau_{ww}}{\beta})\Gamma(\frac{1}{\beta})$ C. P. Lindsey_1980 , where $\Gamma$ is the Euler Gamma function. The present work studies the approach of Laponite suspensions towards kinetic arrest by systematically monitoring the evolution of the average $\alpha$- relaxation time, <$\tau_{ww}$>, with increasing sample waiting time, $t_{w}$, in the presence of several additives and at different suspension temperatures. The diffusive nature of the dynamics even in the presence of the additives is evident from Figures S1 and S2 of Supporting Information.

Rheological measurements were performed using a stress controlled Anton Paar MCR $501$ rheometer. A double gap geometry with a gap of 1.886 mm, an effective length of 40 mm and requiring sample volume of $3.8$ mL was used for dilute suspensions. For dense suspensions, a cone-plate geometry having cone radius $r_{c}$ = 12.491 mm, cone angle $0.979^{\circ}$, measuring gap $d$ = 0.048 mm and requiring sample volume of $0.07$ mL was used. For the rheological measurements, the sample was loaded immediately after preparation. Loading memory effects were erased by shear melting the sample at a high shear rate of 1500 $s^{-1}$ for 3 minutes to achieve a reproducible starting point for all experiments. The increase in the sample waiting time, $t_{w}$, was monitored continuously from the time shear melting was stopped. The temperature of the sample was controlled using a water circulation system (Viscotherm VT2). Silicone oil of viscosity 5cSt was used as a solvent trap oil to avoid water evaporation. The viscoelastic properties of aqueous Laponite suspensions in the presence of different additives and at different suspension temperatures were then investigated by performing oscillatory amplitude sweep experiments. By varying the strain amplitude, $\gamma$, from 0.1% to 500%, the storage modulus, $G^{\prime}$, and loss modulus, $G^{\prime\prime}$, were measured at a constant angular frequency, $\omega$, of 6 rad/s.

Aqueous Laponite suspensions in the presence of dissociating and non-dissociating molecules were visualized using a field-effect scanning electron microscope from Carl Zeiss with an electron beam strength of 5 kV. The samples were loaded in capillary tubes (Capillary Tube Supplies Ltd, UK) with a bore size of 1 mm using capillary flow. The ends of the capillaries were sealed. Samples were kept undisturbed before imaging and were then vitrified using liquid nitrogen slush at temperature -$190^{\circ}$C. The vitrified samples were cut and sublimated for 15 min at -$90^{\circ}$C and then coated with a thin layer of platinum (thickness approximately 1 nm) in vacuum conditions using a cryo-transfer system (PP3000T from Quorum Technologies). The cryo-chamber temperature was kept at -$190^{\circ}$C. Backscattered secondary electrons were used to produce surface images of the samples. ImageJ (Java 1.8.0_172, developed by Wayne Rasband, NIH, US) was used to analyze the cryo-SEM images.

Figure 2 shows the evolution of the slow $\alpha$- relaxation time <$\tau_{ww}$> with waiting time, $t_{w}$, estimated from fits of the DLS intensity autocorrelation data to eq 1, for 12.2 mM (2.8% w/v) aqueous Laponite suspensions (${\bf\color[rgb]{1,0,1}{\pentagon}}$) at 25^∘C in the presence of different concentrations of dissociating and non-dissociating additive molecules in the dispersion medium. We observe that when the concentration of glucose is increased, <$\tau_{ww}$> increases rapidly as waiting time $t_{w}$ is increased (Figure 2a). It is well known that the aging dynamics of aqueous Laponite suspensions is an osmotic pressure driven phenomenon Y_M_Joshi_2007 ; D_Saha_2015 . The enhancement of suspension aging in the presence of glucose, manifested by the rapid time-evolution of $<\tau_{ww}>$ (Figure 2(a)), can be explained by considering the enhanced osmotic pressure gradients that arise due to the formation of tight hydration shells around glucose molecules M_A_Darwish_2014 ; V_Gekas_1998 . On the other hand, when DMF of different concentrations (shown in dark yellow symbols in Figure 2(b)) is added to Laponite suspensions, the divergence of <$\tau_{ww}$> is delayed when compared to Laponite suspensions without any additives (${\bf\color[rgb]{1,0,1}{\pentagon}}$) at the same temperature. The addition of DMF molecules disrupts hydrogen bonds in water, with the water molecules forming hydrogen bonds with the oxygen sites of DMF Y_Lei_2003 ; J_M_M_Cordeiro ; P_M_Geethu_2018 . This reduces the diffusion of bulk water molecules into the intra-gallery spaces of the Laponite tactoids and results in a reduction in the osmotic pressure gradients and a retardation in the observed aging dynamics.

It is seen that when the concentrations of NaCl (shown in blue symbols in Figure 2(c)) and KCl (shown in red symbols in Figure 2(d)) in the suspension are increased, <$\tau_{ww}$> increases very rapidly with waiting time $t_{w}$. The Debye screening lengths of aqueous Laponite suspensions are calculated using the expression: $\kappa^{-1}=(\epsilon_{0}\epsilon_{r}k_{B}T/\Sigma_{i}(z_{i}e)^{2}n_{i})^{1/2}$ Israelachvili_2010 , where $\epsilon_{0}$, $\epsilon_{r}$, $k_{B}$, $z_{i}$, $n_{i}$ and $e$ are the permittivity of free space, relative permittivity, Boltzmann constant, valency of the $i$th ion, total number density of the $i$th ion and electron charge respectively. The details of the calculation for Debye screening lengths are provided in section 1(b) of the SI. It is observed that the Debye screening lengths, $\kappa^{-1}$, of aqueous Laponite suspensions decrease continuously with waiting time which indicates the continuous dissociation of sodium ions from the Laponite faces. The addition of NaCl in the dispersion medium leads to a further decrease in $\kappa^{-1}$ (inset of Figure 2f), indicating an increase in the induced inter-particle electrostatic attractions in the Laponite suspensions, presumably due to OC and HOC associations Samim_2016 of the Laponite platelets. The water molecules form tight hydration shells around Na⁺ J_Cannon_2012 which increases osmotic pressure gradients in the suspension. An increase in both inter-particle electrostatic attractions and osmotic pressure gradients contribute to the observed acceleration in the aging dynamics of Laponite suspensions in the presence of NaCl (Figure 2c). When KCl (widely regarded as a chaotrope and shown in red symbols) is added to the medium, the aging dynamics of the suspensions are seen to accelerate inspite of the chaotropic action of K⁺ on the dispersion structure (Figure 2d). The accelerated aging dynamics is attributed to a decrease in the Debye screening lengths (inset of Figure 2f) due to the participation of K⁺ ions in the EDL around each Laponite particle. This leads us to conclude that in spite of the kosmotropic nature of Na⁺ and chaotropic nature of K⁺ ions, electrostatics is the driving force behind the observed accelerated suspension aging in the presence of dissociating salts such as NaCl and KCl. Clearly, electrostatic interactions dominate over any changes arising from structural modifications in suspensions containing dissociating additives.

The solid lines in Figures 2a-d are fits to the following equation P_Gadige_2017, ; D_Saha_2014, ; D_Saha_2015,

where $D$ is the fragility parameter and $<\tau_{ww}>^{0}$ is the mean $\alpha$- relaxation time when $t_{w}$ $\rightarrow$ 0. The parameter $t_{\alpha}^{\infty}$ is identified as the Vogel time or waiting time at which the suspension achieves a kinetically arrested state D_Saha_2014 . The time-scales $t_{\alpha}^{\infty}$ and $<\tau_{ww}>^{0}$ for aqueous Laponite suspensions with and without additives are extracted by fitting the data to eq 2 and are plotted in Figure 2e. In Figure 2f, we superpose the <$\tau_{ww}$> vs. $t_{w}$ data for all the suspensions (Figures 2a-d) on a universal curve by dividing the horizontal and vertical axes by $t_{\alpha}^{\infty}$ and $<\tau_{ww}>^{0}$ respectively. The self-similar curvatures of the data indicate a common underlying mechanism in the aging of Laponite suspensions with and without additives. The observed increase in $<\tau_{ww}>^{0}$ (shown in dark navy symbols in Figure 2e) and the simultaneous decrease in $t_{\alpha}^{\infty}$ (shown in dark purple symbols in Figure 2e) in the presence of glucose, NaCl and KCl point to the increased confinement of Laponite particles in deep potential energy wells and a rapid acceleration in the aging dynamics. The addition of DMF, on the other hand, resuls in a decrease in $<\tau_{ww}>^{0}$ and an increase in $t_{\alpha}^{\infty}$, indicating shallower potential wells and retarded aging dynamics.

Figure S3 of the SI shows the conductivity of aqueous Laponite suspensions with and without additives. Owing to the uncharged nature of glucose molecules, we see a decrease in the conductivity of the suspension in the presence of glucose. The enhanced aging dynamics of Laponite suspensions in the presence of glucose is therefore understood in terms of enhanced osmotic pressure gradients arising from the formation of hydration shells. The addition of DMF has been shown by us to delay aging (Figure 2(b)). Although DMF is known to be conducting M_Y_Onimisi_2016 , the disruption of water structure upon the addition of DMF in Laponite suspensions clearly dominates over any changes in interparticle electrostatics. Since glucose and DMF molecules do not participate in the EDL, these results establish the importance of structural changes in the dispersion medium in determining the suspension dynamics. EDLs of Laponite suspensions shrink in the presence of NaCl and KCl (inset of Figure 2f). While the present results confirm earlier observations of accelerated aging in the presence of NaCl D_Saha_2015 ; K_Suman_2018 , we note the same effect even in the presence of KCl, a dissociating chaotropic molecule. As discussed earlier, the active participation of K⁺ in the EDL results in the dominance of electrostatic screening over any disruption of medium structure that KCl may cause.

We next perform oscillatory strain amplitude sweep rheology experiments to study the viscoelastic properties of aqueous Laponite suspensions with and without additives. The viscoelastic moduli ($G^{\prime}$ and $G^{\prime\prime}$) of 12.2 mM aqueous Laponite suspensions ($t_{w}$ = 200 min) with different concentrations of additives are plotted as a function of applied strain amplitude in Figures 3a-d. At small values of the strain (LVE regime), the elastic modulus, $G^{\prime}$, and the viscous modulus, $G^{\prime\prime}$, are independent of the applied strain, with $G^{\prime}$ > $G^{\prime\prime}$ for all concentrations of added, glucose, NaCl and KCl and for samples with 50 and 90 mM DMF . The suspensions start yielding with an increase in the applied strain amplitude which is characterized by a monotonic decrease in $G^{\prime}$, while $G^{\prime\prime}$ shows a peak before also decreasing. These are typical features of soft glassy systems with the suspensions showing predominantly fluid-like characteristics indicated by $G^{\prime\prime}$ > $G^{\prime}$ at very high strain amplitudes. In contrast, for Laponite suspensions with very high DMF concentrations (130 mM and 260 mM DMF in Figure 3b), $G"$ > $G^{\prime}$ in the entire strain window, signifying liquid-like behavior.

In Figure 3e, we plot the linear modulus, $G^{\prime}_{l}$ (the magnitudes of $G^{\prime}$ extracted at very low applied strain amplitude $\gamma$, shown in violet symbols), and yield stresses, $\sigma_{y}$ (shown in wine symbols), of aqueous Laponite suspensions with and without additives at $t_{w}$ = 200 min. The yield stresses are calculated from amplitude sweep data following the method proposed by Laurati et al. Laurati_2011 . The details of the analysis and a representative plot are presented in section 1(c) and Figure S4 of the SI. Both $G^{\prime}_{l}$ and $\sigma_{y}$, which estimate the strength of the underlying sample microstructures Samim_2016 , increase with increasing concentrations of glucose, NaCl and KCl, and decrease with increasing DMF concentration. The rheological results are therefore consistent with the DLS results reported earlier. In Figure S5 of the SI, we normalize the storage and loss moduli data measured in oscillatory strain sweep rheological experiments by dividing the moduli at different strain amplitudes with those measured at a strain ($\gamma$ = 0.5%) in the LVE regime G_Y_H_Choong_2013 . As expected, the microscopic dynamics and rheological responses in the limit of low strains collapse well. In contrast, the mechanical responses of Laponite suspensions at high strains are sensitive to the presence of additives.

We perform cryo-SEM experiments to investigate the morphologies of the samples studied here. Figures 4a-e display the microstructures of aqueous Laponite suspensions without and with additives at a waiting time of 24 h. Honeycomb like network structures are seen in all samples, with pore sizes that are sensitive to the additive present in the suspension. We note the existence of holes on the flat surfaces of the network branches, indicating the possibility of overlapping coin configurations (OC) B. Jonsson_2008 of the Laponite platelets Samim_2016 . The magnified cryo-SEM micrographs of the sample microstructures are provided in Figure S6.

We adopt a protocol used earlier Samim_2016 to quantify the porous microstructures by estimating pore areas (data at $t_{w}$ = 24 h is shown in black symbols in Figure 4f) and branch thicknesses (Figure S7). In the analysis of cryo-SEM images, the presence of vitrified water on the structures can lead to the overestimation of the network branch thickness and a simultaneous underestimation of average pore area. However, since the sublimation time (15 min) after cutting the vitrified samples is identical in all the experiments, an equal sublimation-depth is expected for all the samples studied using cryo-SEM. We correlate network morphologies with their mechanical responses obtained in rheological experiments. The linear modulus, $G^{\prime}_{l}$ (magnitude of $G^{\prime}$ of all the samples at $t_{w}$ = 24 h extracted at very low applied strain amplitude $\gamma$ (Figure S8)), are plotted in Figure 4f (shown in red symbols). While the branch thicknesses of the samples remain unchanged in all the samples studied here, the average pore area decreases for Laponite suspensions in the presence of glucose ($\Diamondblack$), NaCl ($\blacktriangle$) and KCl (). Figure 4f displays a simultaneous increase in the elasticity of these samples. In contrast, an increase in average pore area and a simultaneous decrease in elasticity is noted when DMF of concentration 90 mM ($\CIRCLE$) is added to Laponite suspensions (Figure 4f). Many small dangling branches are observed in the walls of the sample microstructures in Figure 4c, which indicates incomplete structure formation and retarded aging dynamics in the suspension. In Figure S9, we show the microstructures of aqueous Laponite suspensions in the presence of all the additives at $t_{w}$ = 200 min. When glucose, NaCl and KCl are added to Laponite suspensions, network structures, with pore areas larger than those observed in the same samples at a longer waiting time ($t_{w}$ = 24 h), are seen (Figure S9f). This indicates the continual aging of the suspension structures. Clearly, aqueous Laponite suspensions age faster in the presence of dissociating molecules such as NaCl and KCl due to the dominance of inter-particle electrostatic interactions over alterations in the dispersion medium structure. The contribution of microstructure dominates only in the case of uncharged molecules, with the kosmotropic glucose and the chaotropic DMF respectively causing acceleration and delay in suspension aging.

Figure 5a plots the dependence of the mean slow $\alpha$- relaxation time, <$\tau_{ww}$>, estimated from fits of intensity autocorrelation functions (eq 1) acquired in DLS experiments, on $t_{w}$, for pre- and post-heated Laponite suspensions. We observe that higher suspension temperatures result in accelerated aging dynamics. This kosmotrope-like effect of temperature on the aging dynamics in a post-heated suspension has been reported in an earlier work D_Saha_2015 . We note that while increasing temperature disrupts hydrogen bonds, there is a simultaneous enhancement in the inter-particle electrostatic attraction, arising due to decrease in the Debye screening length $\kappa^{-1}$ ( Figure 5b, calculated from conductivity measurements plotted in Figure S10 of the SI). At increased temperatures, therefore, electrostatics dominates over disruption in medium structure while determining suspension dynamics. Interestingly, we observe a comparatively more rapid enhancement in the aging dynamics of aqueous Laponite suspensions when temperature of water is raised before the addition of Laponite particles (Figure 5a). The enhanced aging in pre-heated samples when compared to post-heated ones arises from the larger decrease in Debye lengths in the former case (Figure 5(b)) due to an increased dissociation of Na⁺ in the former experimental protocol. The superposition plot of the <$\tau_{ww}$> data (Figure 5c), shows self-similar behavior, thereby revealing an indistinguishable aging mechanism of aqueous Laponite suspensions at all temperatures D_Saha_2015 . The horizontal and vertical shift factors, $t_{\alpha}^{\infty}$ and $<\tau_{ww}>^{0}$ (extracted by fitting the data in Figure 5a to eq 2), are plotted in Figure 5d. The observed decrease in $t_{\alpha}^{\infty}$ (displayed in Figure 5d using blue symbols) with increase in suspension temperature indicates accelerated aging dynamics due to accelerated structural buildup in the suspensions. Lower values of $t_{\alpha}^{\infty}$ (${\bf\color[rgb]{0,0,1}{\blacksquare}}$) in pre-heated suspensions when compared to post-heated ones (${\bf\color[rgb]{0,0,1}{\blacktriangleright}}$) quantitatively verify the presence of accelerated aging dynamics in the former case. While a monotonic decrease in $<\tau_{ww}>^{0}$ with temperature in post-heated suspensions (${\bf\color[rgb]{1,0,0}{\blacktriangleright}}$) indicates the chaotropic effect of temperature at time $t_{w}$ $\rightarrow$ 0, we observe a non-monotonic behavior in $<\tau_{ww}>^{0}$ in pre-heated suspensions (${\bf\color[rgb]{1,0,0}{\blacksquare}}$). A higher value of $<\tau_{ww}>^{0}$ at 15^∘C in the pre-heated suspension reveals the dominance of an initial kosmotropic behavior. In contrast to post-heated suspensions, $<\tau_{ww}>^{0}$ of the pre-heated samples increase with temperature at T > 25^∘C due to rapid acceleration in the aging dynamics under these conditions. The influence of water structure on suspension dynamics at larger $t_{w}$ is minimal in most of these experiments.

Next, oscillatory amplitude sweep rheological experiments are performed to study the viscoelastic behavior of aqueous Laponite suspensions at different temperatures. Figure 6a shows the temperature dependence of the viscoelastic moduli ($G^{\prime}$ and $G^{\prime\prime}$) of 12.2 mM aqueous Laponite suspensions at $t_{w}$ = 200 min as a function of applied strain amplitude, $\gamma$. The linear modulus, $G^{\prime}_{l}$ (olive symbols), and yield stresses, $\sigma_{y}$ (orange symbols), of aqueous Laponite suspensions increase with temperature (Figure 6(b)) which indicate enhanced rates of structure buildup Samim_2016 and accelerated aging dynamics. The higher $G^{\prime}_{l}$ and $\sigma_{y}$ values at higher temperatures in the pre-heated suspensions when compared to post-heated ones confirm faster aging in the former case and are in agreement with the results of the DLS experiments reported in Figure 5. Similar to observations in Laponite suspensions with externally added additives, we note that while the microscopic dynamics and linear rheological responses collapse well at all temperatures (except at 15^∘C for the pre-heated samples (Figure 5(d)) in which the kosmotropic effect dominates), the nonlinear mechanical responses are sensitive to temperature histories over the experimental strain windows (Figure S11 of the SI). We conclude that even though increase in temperature disrupts hydrogen bonds in the dispersion medium, the enhancement in inter-particle electrostatics due to increased dissolution of Na⁺ from the Laponite particles dominates suspension dynamics. While the kosmotrope like effect of temperature has been demonstrated before D_Saha_2015 ; K_Suman_2018 , the differences between pre-heating and post-heading have never been reported in the literature to the best of our knowledge.

Section 1(a) : Diffusive behavior of Laponite suspensions in the presence of different additives are shown in Figures S1 and S2. Section 1(b) : Discussion of Debye screening length calculations. Figure S3 : Plots of conductivity of Laponite suspensions in the presence of different additives. Section 1(C): Details of yield stress calculation, with representative data plotted in Figure S4. Figure S5: Plots of normalized storage and loss moduli vs. applied strain. Section 1(d), Figures S6-S8: Magnified cryo-SEM images, calculation of network branch thicknesses of Laponite suspensions and strain amplitude sweep rheological data in the presence of different additives at waiting time 24h. Figure S9: Magnified cryo-SEM images in the presence of different additives at waiting time 200 min. Section 2, Figures S10-S11: Conductivity measurements vs. sample age and normalized storage and loss moduli data vs. applied strain at different suspension temperatures.