Magnetic control of heterogeneous ice nucleation with nanophase magnetite: Biophysical and agricultural implications

Physical Processes

Magnetite nanoparticles dispersed in water will be subjected to two major physical processes that influence their interaction with the surrounding water molecules. First, Brownian motion, particularly the particle rotations, driven by thermal agitation of the particles will tend to disrupt the surface boundary layer of water molecules and inhibit nucleation. Second, magnetite nanoparticles >30 nm in diameter lie within the single-domain stability field wherein the shape and magneto-crystalline anisotropy energies hold the net moment of the crystal fixed relative to the particle (14). The magnetic moment of the particles will interact with the ambient magnetic field, producing a torque that acts to rotate them into alignment with the field. If this field is moving, this process will contribute to the disruption of the surface layer interacting with the magnetite crystals. The balance between these time constants is given by the extension of the classical Langevin function, applied to the case of uniformly magnetized (single-domain) particles in which the magnetic anisotropy energies are large enough to pin the net magnetic moments to the crystal axes, forcing the particles to rotate; for a thorough treatment, see Coffey and Kalmykov (15).

As we anticipate that many of the crystals of biological magnetite in animal tissues would be bound to membrane systems of some sort, we note that a small force on a biological membrane will cause it to deform with a characteristic time constant of about 0.1 s (16), or a 10-Hz frequency. Hence, at higher frequencies interactions from adjacent cellular structures might damp the rotation. We also note that if these nanometer-scale particles are imbedded in stiff tissues like bone or cartilage, the supporting structures would also block their access to the diffusion of water molecules that would propagate ice nucleation. For both of these reasons, we set the frequency for our initial experiments at 10 Hz.

Rationale for the Field and Frequency Settings Used Experimentally.

To minimize the area on the surface of the magnetite crystals that remains at rest with respect to the aqueous medium, we chose to use a precessing magnetic field, composed of a static vertical component, B_v, and a rotating horizontal component of constant magnitude, B_h. Setting the magnitude of B_v equal to B_h yields a dip angle of the total magnetic vector with respect to horizontal of 45°. A simulation of a particle tracking this applied field is shown in Movie S1, which shows how all points on the surface of the particle have a nonzero magneto-mechanical motion added relative to the surrounding liquid. This is not the case with simple field rotation in a plane or an oscillation along one axis, as under those conditions there are spots on the particle’s rotation axis that have no net motion with respect to the surrounding water, and hence might nucleate ice crystals.

From the above analysis, we chose to use ∼50-nm diameter magnetite particles, field strengths at 0.5, 1, and 1.5 mT, and 10 Hz for the rotating component of the field. Under these conditions, it is easy to show that the electrical fields induced in our balloons are far below the levels known to orient water molecules based on their slight electric dipole moments (see ref. 21 for a recent review).

A detailed description of the experimental setup as modified from Kobayashi et al. (11) is given in SI Materials and Methods and in Fig. S1. Fig. 1 here shows components of the magnetic exposure facility, which consists of three pairs of orthogonally nested square Helmholtz coils capable of producing static and rotating fields of up to 2 mT (20 G). Also shown is a schematic of the wiring diagram, which allows fully active and active-sham experimental modes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Schematic diagram of the set of orthogonal square Helmholtz coils used in this study, and the wiring scheme for control experiments. (A) Nested square Helmholtz coils that are located within a temperature-controlled chamber inside a −80 °C freezer, as described in Fig. S1. The samples being tested for supercooling are shown by the balloon in the center of the coil system. (B and C) Coil wiring scheme using a double-pole, double-throw (DPDT) switch for controlling the active/sham states, following the recommendations for doing blinded experiments in biomagnetic studies (47). This involves the use of a double-wrapped configuration in which two strands of copper magnet wire wrapped in parallel are configured via the DPDT switch to either produce an external magnetic field (B) or a net-zero “sham” field (C), depending on the direction of current in both wires. In the sham mode, the heating and vibration artifacts are the same, but no external magnetic field is generated.

Discussion/Analysis

Our primary result that weak, rotating magnetic fields significantly inhibit ice nucleation in water spiked with nanophase magnetite confirms the hypothesis that this mineral is a powerful ice nucleation agent (11, 12). Comparison with studies of other mineral (7, 10, 24), synthetic (8), and biological agents (9) suggests that it might be among the most efficient of such materials. Additional experiments here also indicate that simple schemes involving magneto-mechanical motion of these particles in suspension, which are designed to “stir up” the water in the vicinity of the crystals beyond that already due to Brownian motion, have a strong ability to inhibit ice nucleation and promote supercooling. These experiments in particular indicate that the ice nucleation process normally happens in and around the boundary layer of water molecules that surround the magnetite mineral surface. In principle, any external action that disturbs this boundary layer ought to promote supercooling.

Our data also confirm the surprising observation that the water in these balloons expands significantly less on freezing than that in which ice nucleates at or near its melting temperature (0 °C). Conditions that promote supercooling, like purification of the water and the magnetic oscillations on magnetite-spiked water, yield this effect. As we noted previously (11), the cause of this volumetric difference is unknown but could possibly arise from trapped gasses being preferentially incorporated as clathrates (rather than forming tiny bubbles) during the rapid freezing jump. Another possibility is that the structurally disordered ice polymorph, I_sd, that apparently cocrystallizes with the hexagonal form (I_h) from supercooled water (25) might expand less. Whatever the mechanism, smaller changes in water volume upon freezing of plant and animal tissues from the supercooled state ought to result in less tissue damage; many studies have shown that cellular damage is minimal from tissues that were supercooled before freezing (see ref. 13 for a discussion). Coupled with a reduction in the needle-like morphology of ice I_h, our data provide a potential biophysical mechanism for past claims of magnetic fields affecting the texture of frozen plant and animal tissues, including food products (21, 26⇓⇓⇓–30).

Although living organisms make nearly 60 different biominerals (31), most of them are in specialized tissues like teeth, bone, or spicules, and most are not known to promote ice crystal nucleation. For many soft tissues, the background levels of biogenic magnetite could provide the largest surface area capable of ice nucleation. Although as yet preliminary, our initial experiments on plant and animal tissues (celery and bovine muscle) show that rotating magnetic fields do indeed promote supercooling, supporting the concept that magnetite dominates the ice crystal nucleation in these materials. It is also interesting to note that our samples of celery did not supercool at the weakest field levels we used (0.5 mT) but did so at the 1.0- and 1.5-mT levels. Past studies of botanical magnetite in plants (the grasses of ref. 32) indicated that those crystals were in the 30-nm size range, rather than the ∼50-nm range used in this study. Hence, stronger fields were presumably needed to compete with Brownian motion. Detailed characterization of the natural size distribution of these magnetite particles, using well-established techniques of rock and mineral magnetism, could be used to predict the optimal magnetic conditions for damage-free cryopreservation. It would also be interesting to determine whether natural selection for survival in frigid environments has lowered the biological production of magnetite in the tissues of some organisms; understanding how such reduction was accomplished might aid in the development of similar traits in the human food supply. However, those implications are beyond the scope of this present work.

Gunders (33) and Gunders et al. (34) estimate that nearly 40% of farm produce in the United States is ultimately lost before it can be consumed, mostly at the production site and in the kitchen. Freezing without tissue damage can reduce these losses, so it is important to identify the physics of heterogeneous ice nucleation and identify the nucleation sites. Our data indicate that magnetite nanocrystals provide one of these important nucleation sites. Nanocrystals of magnetite form through biological processes in a variety of plant (32) and animal (35⇓⇓⇓–39) tissues, including humans (40⇓⇓–43). This magnetite is not just for sensory purposes, as SQUID moment magnetometry reliably detects the presence of parts-per-billion to parts-per-million levels in a variety of tissues, some of which are not known to have sensory neurons [like plant tissues (32) and animal tumors (43)]. Magnetite nanoparticles of exogenous origin can also work their way into tissues (44), as appears to be the case in humans exposed to high levels of industrial pollutants (45, 46). Whatever the source, SQUID moment magnetometry on many of these typical biological tissues indicate that levels of ferromagnetic materials consistent with magnetite in the range of a few parts per billion to a few parts per million is often present in plant and animal tissues. Note that a single 50-nm crystal of magnetite in a typical ∼50-μm eukaryotic cell represents a volume fraction of only 1 ppb, but that particle could well represent the only surface capable of nucleating ice; once nucleated, the entire cell or local assembly of cells would freeze. Hence, nanocrystals of magnetite may ultimately be responsible for nucleating the ice crystals that damage biological tissues during freezing.