This discussion forum focuses on the nature of micro-organisms and other organisms that are ice nucleators (can catalyze freezing of supercooled water) and their role in cloud processes and precipitation.


You are not connected. Please login or register

Sugar and ice: evaluating the interactions between heterogeneous ice nucleators and solutes

View previous topic View next topic Go down  Message [Page 1 of 1]

In our attempt to understand the mechanisms by which ice nucleators catalyze freezing, we biologists have sometimes imagined ice nucleation active proteins as having an attractive force that draws water molecules into the appropriate configuration to form ice. Our simplified model involves water molecules and proteins. But what would happen to the interaction of water molecules and these proteins in the presence of other forces on the water molecules? The forces that can bind water molecules to various degrees are numerous. Gravitation, capillarity and pressure can reduce the potential energy of water. Dissolved materials such as salts or sugars can interact with water through ion-dipole or dipole-dipole attraction and hydrogen bonds. Surface interactions with undissolved materials can also influence water via ion-dipole attraction, van der Waals forces, hydrophobic interactions and hydrogen bonds. A decrease in the boiling point of water with increasing altitude and the freezing-point depression observed for sugar and salt solutions are examples of how these forces act on water.

The recent paper of Thomas Koop and Bernhard Zobrist (reference below) presents an analysis of the influence of solutes on heterogeneous ice nucleation by using approaches from studies on homogeneous nucleation. For biologists, and particularly those of us who have only vague memories of plant or microbial physiology, a quick refresher course on the definition of water activity and water potential will be useful in understanding this paper. One of the objectives of their work is to propose a generic model for estimating the influence of solutes on heterogeneous ice nucleation activity. They compare methods based on the lambda parameter (a parameter whose value changes depending on the nature of the solute) and on water activity. This latter parameter can be estimated directly from the melting point (and they present tables in their paper for this estimation). These authors then use this approach to illustrate how a range of invertebrates balance their investment in decreasing the water activity of their tissues and in forming heterogeneous ice nuclei depending on if they are frost tolerant or if they strive to avoid freezing.

This approach has other applications concerning biology, and in particular in identifying solute conditions that denature heterogeneous ice nuclei. I suppose that this would be revealed by heterogeneous ice nucleation temperatures that were markedly different from those predicted by extrapolation of the water activity vs. ice nucleation curves. In Koop and Zobrist’s paper the behavior of the biological ice nucleator tested (the Snomax formulation of Pseudomonas syringae) suggested that there was no aberrant influence of the solutes used here. I look forward to a future paper for indicators of denaturing effects.


Koop, T., Zobrist, B. 2009. Parameterizations for ice nucleation in biological and atmospheric systems. Phys. Chem. Chem. Phys. 11:10839-10850. http://dx.doi.org/10.1039/b914289d

View user profile

2 more on solute effects on ice nucleation on Sun Nov 29, 2009 8:36 pm

The 24-11-2009 post by Cindy Morris, and the publication by Koop and Zobrist (2009; KZ09) that she discusses prompt me to add a few comments. First my understanding of the issue, and then some data; underscoring what has been said and adding a bit to the horizon.

The effects of solutes on ice nucleation is basically through the same changes in water structure as the depression of the equilibrium melting point (Tmelt) of the solution in comparison with pure water. This is perhaps fairly obvious at first glance but it really isn't. It says a great deal about what controls nucleation, that is the energy of embryo formation. The indication is that, at the equilibrium melting temperature, the solute makes it easier - requiring less thermal energy of molecular motion - for water molecules to move from ice to liquid, and conversely making it less likely for water molecules to attach to existing ice surfaces from the liquid. Furthermore, the same added difficulty seems to apply to the first formation of an ice embryo (i.e. nucleation). Yet further, the effect applies to homogeneous nucleation as well as to heterogeneous nucleation.

On closer examination the beautiful simplicity of the story gets a bit spoiled. That is what leads to the extensive literature that now exists on the subject. At least two complications have to be dealt with.

(In the following, the symbol T is used for temperature with self-explanatory subscripts and ΔT designates changes induced by the presence of solutes.)

First, Tmelt is a unique value for any solute and given concentration. The simple description given above about the equality of influence on the melting point and on nucleation very likely holds at Tmelt quite reliably although it can't be tested since it hasn't been possible yet to observe heterogenous nucleation with near-zero supercooling. So, we have consider what happens at temperatures at which nucleation actually occurs, several degrees or many tens of degrees lower than Tmelt. It is on this issue that water activity, aw, becomes so useful, as has been shown by Koop et al. (2000). But it is not a perfect substitute. Now a bridge has to be made between the energetics of water molecules escaping from the liquid solution to the vapor and the energetics of liquid to solid transition. That bridge remains theoretically somewhat elusive, but experiments for homogeneous nucleation provide a relatively simple result that can be expressed alternatively as ΔThomog being proportional to ΔTmelt with a constant of proportionality near two, but varying with solute type, or as a translation of the Tmelt((aw) curve by 0.3 in aw independent of solute type. The meaning of the latter is easier to see in diagrams (Fig 1b of Koop et al. 2000 and Fig 1 of KZ09). To my knowledge, the numerical constants in these two formulations have to be seen as empirical results, not as theoretical predictions, but that is not surprising in view of the uncertainties of nucleation theory in general.

Second, heterogeneous nucleation in solutions adds further twists to the story. Most importantly, as shown in KZ09, the proportionality ΔThetero ~ λ (ΔTmelt) has λ -values that differ from the value for the homogeneous case, vary quite widely with solute type and, more disturbingly, with the type of ice nucleating substance. In terms of water activity, the dependence on solute type can be eliminated but the variations with nucleating substance remain.
KZ09 present data for ATD (mineral dust) and for Snowmax (P. Syr.) and for sulfates, glucose and PEG with solute concentrations of up to 5 mol/L. For Snowmax the shift in aw is 0.09 and for ATD it is 0.21. Both are smaller than for homogeneous nucleation and there is monotonic trend evident between nucleation temperatures and the shifts. As they point out, the results at low solute concentrations differ some, but within experimental errors, from the overall trend. This is noteworthy when combined with the results of Wilson and Haymet (2009). It is also worth noting what the actual freezing temperatures were (in pure water and as roughly read from the published figures): -23° C for ATD, -11° C for P.Syr. in the KZ09 experiments with micrometer-sized droplets in emulsion; -12 to -14° C with a sand grain in the WH09 experiments using 200-400 μL samples.

These new results prompted me to turn back to work a graduate student and I did nearly 35 years ago (Reischel and Vali 1975; RV75) and see how they can be interpreted now. We expressed the influence of a solute in terms of "true supercooling" (TS) which is the difference between the melting temperature and the nucleation temperature. Measurements were made with the drop-freezing technique (10 μL drops) using four different ice nucleating samples and a large number of soluble salts. Results for leaf-derived nuclei (LDN), later shown to have P. syr. as the active agent, are shown below. The right-hand ordinate shows the change in TS due to the addition of solute to the sample. The value of TS in the graph is for 90% of the drops frozen. We were interested in low solute concentrations in order to look at possible effects in tropospheric clouds, so measurements were made at 0.01, 0.1 and 1 mol/L concentrations.

http://i97.servimg.com/u/f97/14/57/91/08/r-v_ld13.jpg

The freezing temperatures of the LDN suspension without solute were in the range -5 to -11°C yielding -9°C for 90% of drops frozen, as shown at the left end of the abscissa. The addition of solutes produced only small changes in TS. In these graphs perfectly horizontal lines would signify λ=1. Reading from the graph, most easily at 1 molal, it appears that TS increased by amounts that are less than ΔTmelt for each salt concentration, so that λ < 2 is indicated. Thus, at this level of scrutiny it can be said that these old results are compatible with the KZ09 results of λ ~ 1.2 for Snowmax with inorganic solutes. Temperature measurements in these experiments were accurate to around 0.1°C and the 90%-values are statistically significant to perhaps double that. Many of the solutes are seen to produce practically identical effects. However, there are some notable differences beyond that. Interestingly, the Li compounds produced no changes in TS at all. The effects of KCl and NH4Br are in the other direction showing larger depressions of TS than the other salts.


The RV75 experiments with AgI nuclei, which had freezing temperatures similar to the LDN sample, showed even more examples of large deviations from constant λ patterns, both upward and downward, and to greater extents than those seen with LDN. The most consistent of the effects for AgI were increases with NH4 salts, even at 0.01 and 0.1 molal concentration for NH4I. The same salts produced consistent and large increases (up to 4°C) with kaolin (freezing temperatures near -20°C). With kaolin, the most extreme effect was a decrease in TS by 12°C due the addition of LiI.

http://i97.servimg.com/u/f97/14/57/91/08/r-v_kl10.jpg

Without getting further into details, the point that can be made based on the RV75 data is that significant deviations from patterns consistent with constant λ-values can occur for specific nucleant/salt combinations. These deviations would also show up in terms of aw analyses. What this means is that the physics of bulk water and of the water/solute interactions do not contain the whole story. Heterogeneous nucleation being determined by specific surface characteristics on near-molecular scale, and since surfaces can be scenes of a large variety of events due to the presence of dissolved molecules (ions), that conclusion is not hard to accept. How much more specific the description can become is a good question. Whether any of this is likely to be relevant to biological systems can't be known without further tests; however, it is clear that one has to be prepared for surprises.[img][/img]

View user profile

View previous topic View next topic Back to top  Message [Page 1 of 1]

Permissions in this forum:
You cannot reply to topics in this forum