Water Freezes Differently on Positively and Negatively Charged Surfaces of Pyroelectric Materials

See allHide authors and affiliations

Science  05 Feb 2010:
Vol. 327, Issue 5966, pp. 672-675
DOI: 10.1126/science.1178085


Although ice melts and water freezes under equilibrium conditions at 0°C, water can be supercooled under homogeneous conditions in a clean environment down to –40°C without freezing. The influence of the electric field on the freezing temperature of supercooled water (electrofreezing) is of topical importance in the living and inanimate worlds. We report that positively charged surfaces of pyroelectric LiTaO3 crystals and SrTiO3 thin films promote ice nucleation, whereas the same surfaces when negatively charged reduce the freezing temperature. Accordingly, droplets of water cooled down on a negatively charged LiTaO3 surface and remaining liquid at –11°C freeze immediately when this surface is heated to –8°C, as a result of the replacement of the negative surface charge by a positive one. Furthermore, powder x-ray diffraction studies demonstrated that the freezing on the positively charged surface starts at the solid/water interface, whereas on a negatively charged surface, ice nucleation starts at the air/water interface.

The ability to control the freezing temperature of supercooled water with auxiliaries, which promote or suppress ice nucleation, provides a critical factor in a variety of areas such as the survival of ectothermic animals, cryopreservation of cells and tissues, prevention of the freezing of crops, cloud seeding, and snowmaking, to mention but a few (1). There are a number of studies, dating back to 1861 (2, 3), indicating that the local electric field near charged surfaces may enhance the freezing of supercooled water (SCW), so-called electrofreezing. This effect is attributed to the ability of the electric field to induce the formation of icelike nuclei. Electrofreezing of ice has been reported near charged metallic electrodes (4, 5) or dielectric surfaces charged by mechanical friction (6, 7), or within the crevices of polar crystals (8). X-ray specular reflectivity measurements (9) and vibrational spectroscopic methods such as sum-frequency generation (SFG) (10, 11), supported by statistical computer simulation studies (1216), have proposed that water molecules near surfaces in general, and near charged surfaces in particular, form clusters that are structurally different from those present in bulk water.

Studies performed with charged metallic electrodes cannot isolate the net effect of the electric field because SCW freezes at noncharged metallic surfaces close to the melting point as a result of the binding of the water molecules to highly conductive surfaces, epitaxy, and the mirror charge effect (4, 5, 17). A way to isolate the influence of the electric field is to use pyroelectric materials, because they are insulators but the charge on their surface can be manipulated. Pyroelectricity is an inherent property of polar materials. Upon cooling or heating, the surfaces of a pyroelectric plate cut perpendicular to the polar axis develop equal but opposite charges (Fig. 1A). The sign of the charge at a given surface is defined by the sense of polarity of the crystal and can be either positive or negative, depending on whether the material is cooled down or warmed up. The charge induced by a change in temperature dissipates within a few minutes because of leakage through and/or around the plate. However, as long as the temperature continues to vary, the charge at the surface persists.

Fig. 1

(A) A pyroelectric plate without compensating (depolarization) charges. The electric field induced by the spontaneous polarization (surface charge) (shown by gray arrows) is confined to the interior of the crystal. (B) At equilibrium, the spontaneous polarization is compensated for by an external (depolarization) charge. A change in temperature changes the spontaneous polarization and induces an electric field inside the crystal [similar to (A)]. After a few minutes, the excess charge dissipates and the equilibrium reestablishes itself. (C) A change in temperature produces an external electric field if a pyroelectric plate is placed in a metal cylinder, because the excess charge of the bottom side redistributes immediately, whereas the charge at the top side requires a much longer equilibration time. (D) Projection of the packing arrangement of a pyroelectric LiTaO3 crystal parallel to the polar axis (z axis) on the (100) plane. The direction (001), called Z+ (up), and the direction Embedded Image, called Z (down), are not equivalent. Upon cooling, these surfaces develop positive and negative charges, respectively.

An unsupported pyroelectric plate with induced surface charges can be viewed as a parallel plate capacitor in which the electric field is confined to the plate interior (Fig. 1, A and B). However, if one of the surfaces is connected to a conductor, then its charge will be redistributed immediately and the direction of the electric field will depend on the conductor’s configuration. For the setup depicted in Fig. 1C, the top surface of the plate becomes charged with respect to the walls of the container, with the electric field perpendicular to the surface. The charge density σ, generated at the surface as a result of some temperature change (for instance, ΔT=10K), is given by σ=α×ΔT=2.3×107C/cm2 (C, coulomb), where α=2.3×108C/(cm2×K) (K, kelvin) is the pyroelectric coefficient of LiTaO3 (Fig. 1C). Although, only a very small fraction of this charge, f<<1, is redistributed to the external conductor, the induced electric field can be very strong. For instance, if f=10% (as estimated for our setup), the electric field is about Ef×σ/(ε×ε0)=320kV/cm (18), where ε0 is the dielectric permittivity of vacuum and ε=81 is the dielectric permittivity of water. Because water has a finite conductivity, the electric field decays (is screened) over ≈0.8 μm (the Debye length for pure water). Because this distance is much larger than the size of a water molecule, the electric field generated in the vicinity of a pyroelectric surface is expected to align the polar water molecules and assemble them in supramolecular clusters. The external electric field can be switched on and off by controlling the charge flow to the container. Accordingly, the effect of the electric field on the freezing of SCW can be compared with and without the field in either direction on the same surface. Here we report on a startling difference in the freezing temperature of SCW between the positively and negatively charged surfaces.

Four types of specimens (four of each type, 8 × 8 mm) were prepared from the same (001) 3-in-diameter, 0.5-mm-thick LiTaO3 wafer (the so-called Z-cut wafer). The specimens had either (001) or (001¯) surface (Fig. 1D) left uncoated and the opposite surface coated either by 200 nm of Cr (henceforth called Zfield+ and Zfield, respectively), or by 500 nm of Al2O3 (henceforth called Znofield+ and Znofield, respectively). The specimens were placed, exposing the uncoated surface, in a copper cylinder with freshly polished inner surfaces mounted on a temperature-controlled (±0.1°C) stage. The presence of the Cr layer at the bottom surface of the Zfield+ and Zfield specimens ensured good contact with the polished copper and facilitated charge distribution, whereas the insulating layer of Al2O3 obstructed charge flow. Consequently, Zfield+ and Zfield specimens developed an electric field at their top surface, whereas Znofield+ and Znofield specimens did not (Fig. 1C). The specimens were cooled in a room with 45 ± 5% humidity at the rate of 2°C/min down from +24°C until the condensation of small (≈0.1-mm) droplets was achieved. Further cooling froze the water, and the freezing temperature was detected (±0.5°C) by observing the droplets with an optical microscope and by detecting the heat released during freezing (19). To ensure reproducibility, before each experiment the top surface of the specimens was cleaned by low-pressure (0.5-mbar) oxygen plasma (see the supporting online material for experimental details) (20). The orientation of the ice crystals was deduced from the x-ray diffraction acquired with a Rigaku ULTIMA III diffractometer. The experiments were repeated and reproduced more than 20 times with each specimen.

Water droplets freeze at –12.5° ± 3°C at (001) and (001¯) surfaces without an electric field (specimens Znofield+ and Znofield). In contrast, the freezing temperatures on the specimens that develop an electric field (Zfield+ and Zfield) differ dramatically from each other and from the surfaces without field. Water droplets freeze at –7° ± 1°C at the positively charged surface (Zfield+ specimens), whereas at the negatively charged surface (specimens Zfield), water freezes at –18° ± 1°C, but the freezing starts at the air/water interface, at which the electric field is much smaller than at the solid/water interface. Furthermore, if the water droplets are kept for 3 to 10 min at –11°C on a Zfield specimen, they do not freeze. However, the 3- to 10-min time period was found experimentally to be sufficient for the negative charge to dissipate. Thereafter, heating of the specimen (by 2°C/min) from –11° to –8°C produces a small but positive charge at the surface, which induces immediate freezing. Thus, on a pyroelectric surface of Zfield specimens, water freezing can be induced by heating, because it creates a positive charge at the (001¯) surface of LiTaO3. These findings indicate that a positively charged surface promotes water freezing, whereas a negatively charged surface retards it, irrespective of whether the crystallization occurs on the (001¯)or (001) surface.

To provide additional insight, we measured the specular x-ray diffraction (XRD) patterns of the ice polycrystals (21). The XRD patterns for Znofield+ and Znofield are almost indistinguishable (Fig. 2A), which implies an absence of epitaxial effects specific to either surface. The intensity of the (002) diffraction peak is at least five times larger than the (100), (101), and (102) peaks. Furthermore, the rocking curve of the dominant (002) peak has a maximum around the inclination angle ϕ=θθBragg=0 (Fig. 2F) (22), which indicates that the crystallites are aligned with the (002) direction (the c axis of ice) perpendicular to the surface. This strongly suggests that the ice nucleation started at the LiTaO3 crystal surface, with the basal (001) face of the ice crystallite attached to the solid surface.

Fig. 2

Powder diffraction patterns of the ice crystals on LiTaO3 surfaces. The freezing points of SCW are indicated on the graphs. (A) to (D) show θ – 2θ scans with the inclination angle ϕ = 0°. (A) Embedded Image and Embedded Image specimens. (B) Embedded Image specimen. (C) Embedded Image specimen. (D) The Embedded Image specimen was kept at –11°C for 10 min without water freezing, and then heating (2°C/min) from –11° to –8°C resulted in freezing. (E) The standard powder diffraction pattern of ice (23). (F) Rocking curve of the Embedded Image specimen of the (002) peak. cps, counts per second. (G) θ – 2θ scan of the Embedded Image specimen with the inclination angle ϕ = 7°C. (H) Rocking curve of the Embedded Image specimen of the (002) peak. (I) Rocking curve of the Embedded Image specimen of the (002) peak. (J) Rocking curve of the Embedded Image specimen of the (100) peak.

As anticipated, the XRD data (Fig. 2, B and C) show that the orientations of the ice crystals produced on the Zfield+ and Zfield specimens are completely different from each other and from those of the uncharged surfaces. The (100) and the (002) diffraction peaks are dominant in the case of Zfield+ (Fig. 2B). The θ2θ XRD scans of the crystals do not show significant differences as a function of the inclination angle ϕ (ϕ = 0°, Fig. 2B, and ϕ = 7°, Fig. 2G), and the XRD patterns are quite close to those of a randomly oriented set of ice crystals (23)(Fig. 2E). In addition, the intensity of the (002) diffraction peak does not change significantly as a function of the inclination angle (Fig. 2I), which supports the conclusion drawn from Fig. 2, B and G, that most of the crystals on the Zfield+ surface are randomly oriented. However, the rocking curve of the (100) peak (Fig. 2J) indicates that there is a small population of crystals that are almost perfectly oriented with the (100) direction perpendicular to the surface (Fig. 2J). In addition, there is a range of inclination angles ϕ=4...+2°, with no crystals having this orientation. The presence of such an isolated population of crystallites suggests that this specific population was nucleated in the vicinity of a flat surface that served as a template. The only flat surface of the droplet is that of the LiTaO3 crystal. The exact mechanism of the nucleation in this regime is very complicated to discern from these XRD studies. In particular, a number of theoretical studies predict that below 20°C, an electric field may arrange water molecules into cubic ice (Ic)–like tri-layers (24) that subsequently convert into the hexagonal (Ih) stable polymorph (25, 26). Nevertheless, irrespective of these complications, the crystallographic phase of the primary nucleating layer becomes an efficient source for the secondary nucleation, which agrees with the almost random distribution of ice crystals on the Zfield+ surface (Fig. 2, B and G).

The XRD patterns of the ice crystals grown at the negatively charged surface of the Zfield specimens exhibit a very prominent feature indicating the source of their nucleation (Fig. 2C). The (002) diffraction peak is much wider but does not show a skew (Fig. 2C). Furthermore, the rocking curve of this peak shows two broad maxima (Fig. 2H), which implies that the nucleation took place on a curved surface that was close but not parallel to the surface of the specimen; that is, at the air/water interface (Fig. 3C). Consequently, one can infer that the nucleation occurs neither in the vicinity of the negatively charged LiTaO3 surface nor in the bulk of the droplets, where the rate of nucleation is 1010 times slower in comparison to that at the air/water interface (27). The XRD pattern of the ice crystals produced by the positively charged surface of the 001¯ face (the charge-reversal experiment, Fig. 2D) indicates that the orientations of the crystals are almost random and are very close to those found when the crystals were grown on the Zfield+ specimen, which supports the conclusion that a positive charge at the LiTaO3 surface strongly facilitates ice nucleation.

Fig. 3

Optical microscopy images of water drops condensed on 7 × 12 mm amorphous (top) and quasi-amorphous (bottom) films of SrTiO3 (100 nm thickness grown on Si) at various temperatures during cooling as follows: (A) –4°C, (B) –11°C, and (C) –12°C. Water freezes at –4°C on the quasi-amorphous (pyroelectric) film and at –12°C on the amorphous (nonpolar) film. The vertical dimension of all images is 1 mm. The space between the samples is an air gap.

Similar differences in the temperature of ice nucleation are found when water freezing is performed on positively and negatively charged quasi-amorphous pyroelectric thin films. Recently, some of us reported that if amorphous films of SrTiO3, BaTiO3, and BaZrO3 deposited on a silicon or silicon oxide substrate are pulled through a steep temperature gradient, they do not crystallize but form a quasi-amorphous pyroelectric phase (28). Transformation of the amorphous phase into the quasi-amorphous phase does not cause a change of composition, and the surface roughness of the quasi-amorphous films is even smaller [4 Å/(1 × 1 μm2)] than that of the amorphous films from which they were prepared [8 Å/(1 × 1 μm2)] (2831). The SCW freezing experiments were performed with the following four types of 60-nm-thick pyroelectric films of SrTiO3 (7 × 12 mm): (i) Those deposited on a bare, highly conductive (degenerate), (001) Si substrate (specimens QAfield+). These specimens develop positive charge on the surface of SrTiO3 upon cooling. (ii) Those deposited on a 100-nm-thick layer of conductive buffer (SrRuO3) on highly conductive Si (specimens QAfield) (32). These specimens develop negative charge on the surface of SrTiO3 upon cooling. (iii) Those deposited on a 50-nm-thick insulating SiO2 layer on Si (specimens QAnofield+). (iv) Those SrTiO3 films on highly conductive Si that were intentionally left amorphous and nonpyroelectric (specimen Anofield). The specimens QAnofield+ and Anofield did not develop charge at their surface in response to temperature change.

The QAfield+,QAfield, QAnofield+, and Anofield films exhibited substantial differences in their freezing temperatures but they were similar to those of LaTiO3. Water droplets froze at –4° ± 1°C on the QAfield+ specimens, whereas on both QAnofield+ and Anofield films, water froze at –12° ± 4°C. On the QAfield specimens, water froze only at –19° ± 3°C. The monitoring of ice nucleation on the QAfield+ and on the amorphous Anofield is shown in Fig. 3.

Our data lead to a few obvious but very important considerations. Water molecules orient their oxygen atoms with their lone electron pairs toward a positively charged surface, whereas water molecules orient their hydrogen atoms toward a negatively charged surface (9, 10). Because the electron density and geometry of the lone electron pairs of oxygen are very different from those of the two hydrogens, one should anticipate that the water molecules self-assemble and interact differently with the surfaces of opposite charge. It still remains an open question how the different orientation of the water molecules near such surfaces may promote or suppress the formation of the cubic or hexagonal ice–like nuclei.

Supporting Online Material

Materials and Methods

References and Notes

  1. The early studies were reviewed in (6).
  2. The fraction of the charge transferred to the conductor is defined by the equivalent capacitances of the pyroelectric crystal and the conductor.
  3. The setup has a relatively small thermal mass (<30 J/K) and is constantly cooled by a cold finger with liquid nitrogen. The control over the temperature is achieved by a computer-driven electrical heater. Because of the small thermal mass of the setup, freezing of the droplets releases a sufficient amount of heat that the temperature versus time dependence deviates from the programmed 2C°/min and shows a constant temperature for 3 to 7 s. During this period, the power consumption of the heater drops by 10 to 15%. Recording both the temperature versus time and the power consumption versus time permits very accurate determination of the freezing point.
  4. The contact angle at room temperature was 7° ± 1° for all LiTaO3 samples (without surface charge).
  5. After nucleation, the droplet converts into a mixture of water and ice at 0°C, which is in contact with the surface of the LiTaO3 crystal through which the heat is taken away. At 0°C, the formation of new nuclei may occur only on the growing ice crystals (secondary nucleation), whereas nucleation at the surface of the LiTaO3 at 0°C does not occur at all; otherwise the supercooling would not be possible to achieve. Therefore, the ice crystals grown on the crystal surface are expected to provide information about the primary nuclei.
  6. The inclination angle ϕ is the angle between the surface normal and the x-ray scattering vector in a θ2θ scan.
  7. International Centre for Diffraction Data powder diffraction file no. 00-016-0687.
  8. Johari (33) proposed, on thermodynamic consideration, that cubic ice–like clusters with a radius smaller than 15 nm and flat ice films 10 nm thick are more stable than the corresponding hexagonal-like counterpart.
  9. The authors acknowledge the Israeli Science Foundation, the Israel Ministry of Science and the Nancy and Stephen Grand Research Center for Sensors and Security for funding this research. M.L. acknowledges the European COST 0703 on system chemistry, the Helen and Milton Kimmelman Center for Biomolecular Structure and Assembly, and the G. M. J. Schmidt Minerva center. The authors thank I. Weissbuch and I. Feldman of the Weizmann Institute for fruitful discussions. This research is made possible in part by the generosity of the Harold Perlman Family.
View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article