betaFeedback

WorksPhysics and Astronomy

Deformation of hydrogel during freezing

Abstract

We conducted experiments to directionally freeze agar hydrogel drops with different polymer concentrations, on a copper substrate maintained at low temperature. Unlike the water droplet studied in the literature, where the liquid part can reorganize during freezing, our droplets show a distinct deformation. This is due to the presence of the viscoelastic solid polymer matrix, which prevents deformation of the unfrozen part, leading mainly to elongation in the direction of the temperature gradient. We then derive a model describing this deformation. It assumes a flat freezing front, a change in volume at the front resulting from the Stefan condition, and a remaining volume to be frozen identical in shape to the initial one. The latter is in good agreement with experimental observations as long as water fluxes in the porous hydrogel remain negligible. It is also conclusive for different object shapes, as the hydrogel can be freely molded without a container.

We can quantify these differences in shape by plotting the relative radius change along the surface for both drops in Figure 1(b) and (c), as a function of the normalized angle θ~=θ−180180−θc\tilde{\theta} = \frac{\theta-180}{180-\theta_c} (inset Figure 1(d)) ranging between -1 and 1. θc\theta_c corresponds to the left and right bottom limits of the drop.

Figure 1. (a) Schematics of the experimental setup. (b) Snapshots of a freezing drop of mixture of water and fluorescein (0.5 g/L) on a cold substrate at Ts=−11∘CT_s = -11^\circ\mathrm{C}. (c) Snapshots of a 4% freezing hydrogel drop on a cold substrate at Ts=−37∘CT_s = -37^\circ\mathrm{C}. The initial shape is represented with a dashed blue line. (d) Relative change in the local radius between the totally frozen drop (rendr_{\mathrm{end}}) and the initial drop (r0r_0) as a function of a normalized angle θ~=θ−180180−θc\tilde{\theta} = \frac{\theta- 180}{180-\theta_c}. The geometrical parameters are described in the inset. Raw data are plotted as points for water (blue) and hydrogel (red) and the curves are obtained from a low-pass filter smoothing. The black dashed curve corresponds to the analytical model for the hydrogel drop (eq. (2)).

The water drop (blue curve, Figure 1(d)) seems to deform uniformly with an overall deformation along the surface lying between 5% and 10%. This confirms an anisotropic distribution of the volume change in the case of the water drop. However, for the hydrogel drop (red curve, Figure 1(d)), we find negative values of the deformation for θ~<−0.5\tilde{\theta} < -0.5 and θ~>0.5\tilde{\theta} > 0.5. This means that the hydrogel contracts close to its area of contact with the copper substrate. Moreover, a sharp peak is observed at the top of the hydrogel drop for θ~=0\tilde{\theta} = 0 with around 15% change in local radius, thus highlighting that a strong expansion is localized at the apex after freezing. In order to assess globally the effect of the agar matrix on the solidification process, we compute the overall volume increase for all the experiments.

This volume increase, vcv_c, is presented on Figure 2 as a function of the polymer concentration, for three different substrate temperatures. It is derived from the integral of curves such as the one presented in Figure 1(d). The black line represents the expected value of 9.2% for the volume increase when water is changed to ice. The first points at C=0%C = 0\% were obtained with pure water drops and agree with the expected value.

Figure 2. Mean volume change vcv_c for different polymer concentrations at different substrate temperatures. The black line represents the theoretical value for pure water (vc=9.2%v_c = 9.2\%).

When the concentration of polymers in the hydrogel is increased, the volume change due to freezing of the sample decreases. Indeed, for a 5% hydrogel, around 3% in volume change is lost. We also find that for concentrations above 1%, the volume change is increasing when the substrate temperature decreases: the dark blue points (-40°C) consistently show a greater expansion than samples at higher temperatures. This decrease in volume change for hydrogels could be linked to the volume of concentrated polymer solution that cannot freeze. First, the presence of the polymer destabilizes the freezing front in crystals aligned in the direction of the temperature gradient [19]. Between these crystals an interstitial liquid, composed of water highly concentrated in polymer, is present [20, 21, 22]. The increasing concentration of solute induces a depression of the freezing point [23], ultimately preventing the freezing and explaining why the volume change in the case of a hydrogel is lower than for water.

Attempts to predict the dependence of the volume change with both the temperature and the concentration theoretically are presented in Supp. Mat. Either based on the Clausius-Clapeyron relation or on empirical laws [24, 25, 26], the predictions give a reasonable trend but the agreement failed to be quantitative. We will thus use the experimental points of Figure 2 as a reference for the rest of the study.

We now wish to investigate how this global volume variation is related to a local vertical expansion for a frozen hydrogel drop. We therefore measure the initial and final height (H0H_0 and HexpH_{\mathrm{exp}} respectively) for hydrogel drops of various concentrations and temperatures. We can then plot the ratio between the drop’s experimental final height and H0(1+vc(Ts,C))H_0(1+v_c(T_s,C)), the final height of a droplet if the volume change were purely unidirectional (Figure 3(a)), as a function of the hydrogel concentration. Here, vc(Ts,C)v_c(T_s,C) is the mean volume change for a given temperature and concentration measured in Figure 2.

Figure 3. (a) Hexp/(H0(vc+1))H_{\mathrm{exp}}/(H_0(v_c+1)) for different polymer concentrations, different temperatures and different sizes. Blue points are colored accordingly to the temperature and black ones are the mean value for each concentration. The size of the blue points is linked to the drop radius. The gray area delimits experiments where a significant water release is observed and the model is expected to fail. The horizontal black line corresponds to the situation where the volume change is only vertical. In this case we usually observe frozen shapes such as the one presented in inset (upper left picture with a 0.5% drop on a substrate at Ts=−11∘CT_s=-11^\circ\mathrm{C}). (b) Schematics of the model: the liquid part is in light gray and the frozen part in dark gray. The contact area is of radius 2Rc2R_c and the initial shape is described by the function R0(z)R_0(z). The freezing front is at height h(t)h(t) and has a radius R(t)R(t). (c) Frozen agar drop on a substrate at Ts=−37∘CT_s=-37^\circ\mathrm{C} and C=4%C=4\%. The blue dashed line corresponds to the initial shape and the yellow one its circle approximation. The red line represents the final theoretical contour for hydrogel (equation (2)).

The blue data points represent the experimental data at varying temperature and the black ones the mean values for each concentration. The size of the dot is proportional to the radius of the drop RdR_d. Interestingly, for C>2%C>2\% the height change is close to the total volume change (HexpH0=vc+1\frac{H_{\mathrm{exp}}}{H_0}=v_c+1), all of the deformation is quantitatively handled in one and only direction: perpendicular to the solidification front. The small error mainly comes from the irregularities at the surface of the drop and from the experimental reproducibility.

For concentrations below 2%, the vertical ascent of the freezing front is below the predicted value, especially for 0.5% and 1%. Indeed, in this case, water is released at the surface of the hydrogel during the freezing process as presented in inset of Figure 3 (and corresponding movie in Supp. Mat.). This water release leads to the formation of bumps at random positions along the surface of the frozen hydrogel, preventing the volume change from being fully vertical. The poroviscoelastic nature of the hydrogel [27, 28, 29] could explain that a stress stemming from the volume increase could trigger a flow of the water in the polymer matrix and its subsequent release along the drop’s surface. The dynamics of the water migration would then depend on the size and porosity of the sample and would be much faster for low-concentrated hydrogels as the pore size would then be larger [30].

The classic theoretical description for the freezing of a drop of water relies on the ability of the liquid part to rearrange its shape through capillary forces [2] (see Supp. Mat. for a model describing the 2D case and comparison with experimental results). However, in the case of a hydrogel, although it almost entirely consists of water, it has a unique behavior with the water being maintained by the polymer matrix and thus preventing rearrangements along the freezing front. We can thus assume that the elastic nature of this soft solid allows the shape of the unfrozen part to remain constant throughout the solidification process. We propose a theoretical model where the deformation is then entirely localized at the front where the local height is subjected to an expansion only in the vertical direction. This deformation spreads towards the top during the freezing process similarly to a propagation front observed in a phase-transforming material such as kirigami shells under strain [31].

First, let us consider a (r,z)(r,z)-surface with a central axis of symmetry parameterized with its distance from the axis RR as a function of the position on this axis zz so that for each zz, RR is defined without ambiguity. The initial shape, of volume v0v_0, is described by the function R0(z)R_0(z), zz varying from 0 to H0H_0.

When a layer of hydrogel of thickness dzdz, initially at the position zz, is changed into ice, the front progresses of dh=(1+vc)dzdh=(1+v_c)dz, as illustrated on the schematics Figure 3(b). This leads to an explicit expression for the final shape of the frozen object R(z)R(z):

R(z)=R0(z1+vc)(1)R(z)=R_0\left(\frac{z}{1+v_c}\right) \tag*{(1)}

with zz now varying from 0 to H0(1+vc)H_0(1+v_c). From this model, we can reconstruct the theoretical frozen shape of any initial hydrogel shape.

For the case of a droplet, modeled as a portion of disk of radius RdR_d in contact with the substrate over a distance 2Rc2R_c (Figure 3(b)), we can derive an analytical expression for the frozen drop. It is given by:

{H(R)=(1+vc)(Rd2−Rc2−Rd2−R2)H(R)=(1+vc)(Rd2−Rc2+Rd2−R2),(2)\begin{cases} H(R)=(1+v_c)\left(\sqrt{R_d^2-R_c^2}-\sqrt{R_d^2-R^2}\right) \\ H(R)=(1+v_c)\left(\sqrt{R_d^2-R_c^2}+\sqrt{R_d^2-R^2}\right), \tag*{(2)} \end{cases}

with HH the height of the object at a given horizontal radius. The first line of equation (2) corresponds to a portion of ellipse below its equator and the second line above. Thus, a liquid truncated disk of radius RdR_d will give a frozen truncated ellipse of horizontal axis RdR_d and vertical axis (1+vc)Rd(1+v_c)R_d. Indeed, our condition on the conservation of the initial shape does not allow any change of the liquid portion in another direction than the freezing front, leading to an unchanged horizontal axis.

Figure 3(c) presents how our model can predict the final shape of an experimental agar drop corresponding to Figure 1(b). The initial drop shape (blue dashed line) is fitted with a circle (yellow line) and the final ellipse is in red. We find a good agreement between theory and experiments, the only measurable error comes from the

lost of circularity of the frozen drop that is reminiscent from the same defect in the initial state. Another way of confronting the model to our experimental results is to plot the variation of radius before and after freezing along the angle θ\theta, as discussed previously in Figure 1(d). The theoretical radius deformation of this particular experiment is plotted in black dashed line. The elliptical shape reproduces both of our observations: a negative change in radius close to the base of the drop and a maximal deformation at its top.

The analytical approach can be further extended to other shapes, such as a rectangle or a triangle. For the rectangular case, described by R0(z)=RcR_0(z)=R_c for 0≤z≤H00\leq z\leq H_0, the frozen shape is given by R(z)=RcR(z)=R_c but for for 0≤z≤H0(1+vc)0\leq z\leq H_0(1+v_c). The initial rectangle is dilated in the vertical direction. Similarly, for an initial triangular shape with a top angle 2α2\alpha and a base 2Rc2R_c, R(z)=Rc−ztan⁡(α)R(z)=R_c-z\tan(\alpha), the frozen shape is also a triangle with a base 2Rc2R_c but with a top angle β=arctan⁡(tan⁡(α)/(1+vc))\beta=\arctan(\tan(\alpha)/(1+v_c)). For a given initial shape with small defects, equation (1) can be used numerically to predict the frozen pattern.

Figure 4 shows different frozen shapes with the initial contour in blue and the theoretical deformation in red. We tested different evolution of horizontal section: constant (Figure 4(a)), constantly decreasing (Figure 4(b)) and non monotonic (Figure 4(c)). Experiments were done on 2% hydrogels with a substrate temperature below −50∘C-50^\circ\mathrm{C}, preventing any water release. As predicted above, we observe that the rectangle on Figure 4(a) has not gained in width but just expanded vertically. This is not the case for the triangle on Figure 4(b) where the final shape, for theory and experiment, is larger than the initial one. As expected, the surface is vertically lifted up but is also keeping is base's length giving a new triangle with a smaller angle at the top. These experimental observations remain valid for experiments with shapes where a brutal change of monotony is imposed on the horizontal section. Figure 4(c) shows a frozen asymmetric hydrogel that has opposite monotony on both sides. On the right, the frozen part has slightly dilated after freezing on the lower half part and then has contracted in the upper half after the change in evolution of the section. The opposite phenomenon is observed on the left side which has two changes in section monotony. Interestingly, most of the deformation is located on the top part of the object, as the vertical deformation is proportional to the height of the object. Overall, our model predicts well the shape of frozen hydrogels.

Figure 4. Frozen hydrogel shapes (2%) (a) Square on a substrate at Ts=−57∘CT_s=-57^\circ\mathrm{C} (b) Triangle on a substrate at Ts=−86∘CT_s=-86^\circ\mathrm{C} (c) Asymmetric shape on a substrate at Ts=−97∘CT_s=-97^\circ\mathrm{C}. Initial shapes are represented by the blue lines and the final theoretical ones by the red lines.

In this study, we present a theoretical model that describes quantitatively the deformation of hydrogels during freezing. Unlike a drop of water that can rearrange its shape during the process, the viscoelastic polymer matrix will maintain the drop’s initial shape, leading to the propagation of a vertical deformation of the object only in the direction of the phase transition. We also investigated the range of validity of our model and showed that it works at low substrate temperature on concentrated hydrogels, as it prevents significant water movements. Finally, we successfully tested our theory on different hydrogel shapes. This problem could be adapted to 3D shapes by removing the assumption of a flat freezing front [32] which would lead to more complex calculation. Furthermore, the water release phenomenon should be investigated, as it could mean that the material undergoes physical properties changes (water content, porosity) during its freezing.

Acknowledgements

SP would like to thank Andreas Carlson who contributed to preliminary stages of this work during a visit at d’Alembert. We thank L. Duchemin for fruitful discussions about the modeling.

References

References

  1. [2]D. M. Anderson, M. G. Worster, and S. H. Davis, The case for a dynamic contact angle in containerless solidification, 163, 329.
  2. [19]W. W. Mullins and R. F. Sekerka, Stability of a Planar Interface During Solidification of a Dilute Binary Alloy, 35, 444.
  3. [20]J. S. Langer, Instabilities and pattern formation in crystal growth, 52, 1.DOI
  4. [21]M. G. Worster, Solidification of an alloy from a cooled boundary, 167, 481.
  5. [22]H. E. Huppert, The fluid mechanics of solidification, 212, 209.DOI
  6. [23]A. Banin and D. M. Anderson, Effects of Salt Concentration Changes During Freezing on the Unfrozen Water Content of Porous Materials, 10, 124.
  7. [24]K. A. Fikiin, Ice content prediction methods during food freezing: A survey of the Eastern European literature, 38, 331.
  8. [25]J. Succar and K. Hayakawa, A Method to Determine Initial Freezing Point of Foods, 55, 1711.
  9. [26]M. S. Rahman, Food Properties Handbook (CRC Press) F9FoNy06qvcC.
  10. [27]D. Caccavo and G. Lamberti, PoroViscoElastic model to describe hydrogels’ behavior, 76, 102.
  11. [28]D. Caccavo, S. Cascone, G. Lamberti, and A. A. Barba, Hydrogels: Experimental characterization and mathematical modelling of their mechanical and diffusive behaviour, 47, 2357.DOI
  12. [29]Y. Hu and Z. Suo, Viscoelasticity and poroelasticity in elastomeric gels, 25, 441.
  13. [30]E. Davies, Y. Huang, J. B. Harper, J. M. Hook, D. S. Thomas, I. M. Burgar, and P. J. Lillford, Dynamics of water in agar gels studied using low and high resolution 1H NMR spectroscopy, 45, 2502.
  14. [31]A. Rafsanjani, L. Jin, B. Deng, and K. Bertoldi, Propagation of pop ups in kirigami shells, 116, 8200.
  15. [32]M. Stiti, G. Castanet, A. Labergue, and F. Lemoine, Icing of a droplet deposited onto a subcooled surface, 159, 120116.DOI

Paper details

vlnc:W_8v6s64ys
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